(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "EHD Research : final report for the year 1968-69"

LIBtCAH* 

mmmmm 



1 MC: 



NPS-57ZI9121A 




United States 
Naval Postgraduate School 




J 





EHD RESEARCH 




FINAL 


REPORT FOR THE 
1968-69 

by 

Oscar Biblarz 


YEAR 


30 December 1969 





This document has been approved for public release and 
sale; its distribution is unlimited. 



FEDDOCS 
! D 208.14/2 
NPS-57ZI9121A 



SoMTWEV CA 93^5101 



C "2- 



NAVAL POSTGRADUATE SCHOOL 
Monterey, California 



Rear Admiral R. W. McNitt, USN 
Superintendent 

ABSTRACT: 



R. F. Rinehart 
Academic Dean 



Interest in electrohydrodynamics (EHD) stems mainly from its 
potential application in electrical power generation. The principle 
is that a stream of insulating fluid, which contains charged particles 
viscously coupled to it, moves charges against an electric field. In 
this fashion, the mechanical power of the flowing medium is converted 
directly into electrical power. EHD generators offer some promise of 
being simple, reliable, compact, and light. 

The present research is concerned with how charged particles can 
be generated in the laboratory with a potentially useful range of sizes, 
of charge, and of number density. It is suggested that refined 
measurement techniques are needed to check on just what is being injected 
into the flow. The effects of turbulence on the EHD process and, 
particularly, on breakdown are being studied. The report discusses 
in some detail the possible role of turbulence on the mean effective 
mobility of charged particles. 

On the experimental side, a laboratory facility has been built 
and then improved by the addition of a larger test section and other 
equipment. Work is proceeding to further develop and refine the 
instrumentation. Two types of injectors have been operated, namely, 
molecular and two-phase and the latter shows potential for efficient 
operation. 

It has been concluded tentatively that turbulence in the carrier 
fluid increases its breakdown potential, and that turbulent air may be 
a suitable medium for the EHD energy conversion process. 

Research plans for the coming year are outlined in the report. 

This task was supported by: Navy Department, Naval Air Systems Command 

AIRTASK No. A34340/551/69RO10O201O 




-^g^-t 



r of Aeronaut! 



Oscar Biblarz 
Assistant Professo 



cs 




r. Bell 
Chairman 
Department of Aeronautics 



Released by: 



C. E. Menneken 

Dean of Research Administration 



NPS-57ZI9121A 
30 December 1969 



DUDLEY KNOX LIBRARY 
NAVAL POSTGRADUATE SCH 
MONTEREY CA 93943-51 Oi 

FOREWORD 



The program on EHD Research was started on September of 1968 at 
the Naval Postgraduate School, Monterey, California. This work was 
sponsored by the Naval Air Systems Command under the technical 
cognizance of Mr. Milton A. Knight. 

The following personnel contributed to the work during the reporting 
year: Assistant Professor Oscar Biblarz (Principal Investigator), 
Assistant Professor D. C. Wooten, and Professor T. H. Cawain. Professor 
Wooten left the program in June of 1969 and Professor Gawain joined in 
July of 1969. Two Master theses we generated during this year and they 
are included as Appendices C and D in this report. 

The help of Mr. Patrick Hickey in the laboratory is gratefully 
acknowledged. 

Since September of 1968, Associate Professor K. E. Woehler 
of the NPS Physics Department has joined the program. 



TABLE OF CONTENTS 

PAGE 

I. INTRODUCTION 1 

A. General 1 

B. Background 2 

C. Method of Approach 5 

II. SURVEY OF WORK IN THE FIELD 7 

III. HARDWARE 1° 

IV. EFFECTS OF TURBULENCE 14 

V. INJECTOR WORK 23 

VI. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS 25 

A. Summary 25 

B. Conclusions and Recommendations 25 

VII. NEW RESEARCH 27 

REFERENCES 37 

APPENDIX A — Idealized Analysis of Basic Energy Conversion 

Process in EHD Flows A-l 

APPENDIX B — Charged Particle Mobility and Diffussion B-l 

APPENDIX C — Ion Injector for Single — and Two-Phase 

Electrogasdynamic Generators (MS Thesis by 

LT(jg) W. M. Ober II) C-l 

APPENDIX D — Molecular-Ion Electrogasdynamic Flow Channel 

(MS Thesis by LT (jg) D. W. Wallace) D-l 

DISTRIBUTION LIST 



ii 



LIST OF FIGURES (MAIN TEXT) 

FIGURE PAGE 

1 Mobility of Water Droplets in Air at 1-Atm 

and 20°C 28 

2 High Free Stream Turbulence EHD Generator 29 

3 "New Test Section" 30 

A Diode Calibration Curve 31 

5 EHD Generator Schematic 32 

6 Air-Water Aerosol Injector 33 

7 Teflon Nozzle Steam Injector 34 

8 Electrical Circuit for Injector Study 35 

9 Maximum Ratio Versus Spacing 36 



iii 



LIST OF TABLES 
TABLE PAGE 

I Schmidt Numbers for Ambipolar Diffusion 17 

II Schmidt Numbers for Space Charge Flow 20 

III Effect of Turbulent Flow on Breakdown of Air Corona ... 22 



iv 



LIST OF SYMBOLS 



A 

&. 



A » Area of duct 

D - Diffusion coefficient, drag force 

D ■ Diffusion coefficient for ions 

Diffusion coefficient for particles 

Diffusion coefficient for electrons 

Space charge diffusion coefficient for ions 

Space charge diffusion coefficient for electrons 

d ■ Cylinder diameter 

E - Electric field 

E - Electric field in the axial direction 
x 

Electric field in the radial direction 

Breakdown field 

Space charge field 
e ■ Charge on an electron 
F » Electrical force 

Ei 

Pressure force 

Viscous force 

v 

G ■ Pressure gradient (-dp/dx) 

h « Height of test section 

I ■ Collector current 
g 

Total current at corona needle 

i 

K ■ Proportionality constant for drag law 
k - Boltzmann's constant 



L 
I 



n 



P 
P 

P 

Q 

Q , 
*T 

R 

R 

( 

r 
r 



r ) 
o max 






T 
U. 



Distance between corona ring and EHD collector 

Typical apparatus dimension, length of duct 

Dimensionless loss parameter 

Specie density 

Number of particles 

Electrical power output 

Fluid power input 

Dimensionless power density parameter 

Dimensionless fluid power 

Total charge of space charge cloud 

Volumetric flow rate for carrier gas 

Total charge per particle 

Particle radius 

Reynolds number based on the diameter 

Radius 

Initial space charge radius 

Maximum value of r 

o 

Schmidt number 

Laminar flow Schmidt number 

Turbulent flow Schmidt number 

Absolute temperature 

Free stream velocity 

Particle velocity 

Corona voltage 

Drift velocity 



vi 



W ■ Width of the test section 
X ■ Flow direction 



x* - Distance for which Tf /U„ - 0.99 

P 

Z ■ Number of electronic charges 



I - Specie flux 

6m ■ Mass of particles 
P 

6TT « Volume of particles 
P 

e » Eddy diffusivity (momentum) 

e - Eddy diffusivity (mass) 
m 

z ■ Permitivity of free space 

n. ■ Efficiency of the Idealized energy converter 

u - Mobility of particles 

V ■ Mobility of particles for 6m 6u -*• 
o 7 v p p 

u m ■ Viscosity of carrier gas 

v ■ Molecular diffusivity (momentum) 

p ■ Electrical power density 

p ■ Fluid power density 

p ■ Particle density 

p m - Carrier gas density 

t - e-folding time for particle velocity charges 



vii 



I. Introduction 

A. General 

Electrohydrodynami.es (EHD) Is the study of the Interaction of a 
non-neutral fluid with an electric field. The fluid contains charged 
particles which are viscously coupled to the surrounding fluid 
particles and, since one polarity prevails, there can be an exchange 
of energy between the fluid and the electric field. The unipolar charged 
particles must be introduced into a highly non-conducting fluid since it 
must support high voltages without any appreciable flow of current. 
The fluid can be either a liquid or a gas, and in the latter case the 
name electrogasdynamics (EGD) is commonly used. Since the injected 
particles must, in general, be quite different from the enveloping 
medium, we speak of EHD flows as two-phase flows (i.e., solid or liquid 
particles suspended in a gas) . 

Interest in EHD stems from its potential application as a 
mechanism for "direct" power generation. Here, the flow energy of an 
insulating fluid is transformed into electrical energy. The principle 
of operation is that of the Van de Graaff generator where the moving 
belt has been replaced by a moving fluid stream. There are three 
basic steps in the EHD generator which are charge injection, charge 
convection by the fluid against the resistance of the elctrical field, 
and charge collection. Power levels of the order of one kilowatt per 
generator unit are anticipated with a variety of energy sources. A 
substantial EHD interaction, if it can be achieved, can also be used 
for purposes other than power generation, such as for propulsion and 
for diagnostic schemes in fluid dynamics. A number of practical 
applications are thus envisioned, but these await a better understanding 
of the underlying phenomena. 

A striking feature of the EHD process is that it permits the 
conversion of fluid kinetic energy into electrical power without the 
need to employ or interpose turbomachinery. This absence of moving 
mechanical parts is a strong practical advantage in some applications 



and can be an important factor in enhancing reliability. EHD, moreover, 
offers advantages over sister schemes such as magnetohydrodynamics (MHD) 
since it generates higher voltages and operates at more moderate 
temperatures. The lower operating temperatures of the carrier fluid 
present significantly fewer material problems when compared to MHD, 
and the absence of a magnetic field in EHD makes for inherently simpler 
equipment and lighter weight. 

During the past decade, there has been considerable interest in 
direct energy conversion devices. The use of EHD generators for this 
purpose has great appeal because of their simplicity of operation, 
i.e., they involve no bulky equipment and need only a very short start- 
up time. Also, they are inherently light weight and, therefore, portable 
Moreover, EHD devices have potential use as smog control devices since 
many of the pollutants can be charged and then precipitated. 

There are at present three major groups in this country who have 

1 o 
interest in EHD generators, viz., the Marks Polarized Company ♦, 

Gourdine Systems Inc-*' , and the Air Force (WPAFB) under the direction 

of Dr. H. von Ohain. 5 ' 6 ' 

B. Background 

It was first thought that EHD is innately an inefficient process. 
More detailed and sophisticated studies have shown, however, that 
theoretical possibilities exist for reaching good performance. Of 
course, the phenomena involved are complex and subtle, and the ultimate 
attainment of these possibilities will require scientific work of high 
order. Before any potential benefits of EHD can be practically 
attained, it is essential to improve our understanding of the basic 
phenomena. Consequently, the present research is aimed primarily at 
the study of such fundamentals. 

There are several problems that research efforts in EHD must 
overcome before producing practical results. For example, because 
the medium has to be highly stressed electrically, one has to contend 
with the numerous ways that nature would like to relieve that stress. 



These include surface leakage through Impurities on insulators, corona 
discharges from sharp metallic surfaces, spark breakdown, etc. 

A major problem in this connection is that of electrical breakdown 
of the carrier gas. Breakdown represents an upper limit of operation 
in the sense that once the fluid begins to generate charge carriers, 
it is no longer an insulator. Breakdown can be most likely attributed 
to the presence of free electrons which gain enough energy from the 
electric field to ionize a molecule upon collision with it. Obviously, 
the geometry of the flow field and the electron scavenging properties 
of the gas are going to affect breakdown. Such scavenging decreases 
the free electron concentration. Additives** such as SFg or CCI2F2 
(Arctron -12) can also be introduced for just this purpose. The geometry 
can have the effect of creating an inhomogeneous field. Consequently 
ionization may take place at lower overall voltage than would otherwise 
be the case, such as in the corona discharge. 

Free electrons always exist in a medium due to a variety of 
natural sources. These electrons need minimum energy for ionization. 
Also there must be a cumulative series of ionizing collisions so that 
there is an avalanche effect and the ionization build-up can overcome 
the losses between the electrodes. The gas density also comes into 
play since there must be enough molecules to ionize (here we get the 
familiar Paschen law^ for spark breakdown). In gases with additives, 
or in gas mixtures 1 , complex interactions can take place which may 
aid or hinder the cumulative processes needed for breakdown. It has 
been found" that, at pressures above atmospheric, the electrode material 
also becomes a significant item in breakdown. There is yet another 
factor which affects breakdown and that is the motion of the gas. Because 
of the strong frictional coupling the charged particles follow the gas 
motion and because of the strong electrical coupling the electrons must 
follow the particles closely (otherwise space charge fields could 
develop which might easily become larger than the external field) . 
Hence, the movement of the gas and, in particular, the degree of turbulence 



of the flow 11 has an effect on the breakdown. This last item will be 
of particular concern to us. 

Another fundamental problem in EHD energy conversion pertains to 
the mobility of the charged particles. Mobility is defined as the 
average drift velocity per unit electric field. It describes how well 
a charged particle couples the long range electric forces to the short 
range forces effective during collisions with surrounding molecules. 
The mobility parameter arises from the observation that in a collision- 
dominated system, particles achieve a terminal drift velocity relative 
to surrounding medium which is proportional to the strength of the 
electric field. This, in essence, is a simplified solution of the 
momentum equation for the charged species. The lower the mobility the 
more highly coupled the particles (and hence the electric field) to the 
fluid. Analyses have shown conclusively that low molecular weight ions 
are too mobile for most cases of interest; larger particles that range 
up to micron dimensions and which carry one or a few charges per 
particle are needed to maintain the mobility at the requisite low 
values. A drift velocity which is smaller than the gas velocity by a 
factor of at least ten is necessary. Low mobility particles not only 
couple the flow field to the electric field but reduce particle 
diffusion to the insulator walls. 

Fig. 1 shows the mobility of water droplets in air as calculated 
by Barreto. * Such theoretical predictions of mobility take into 
account particle size and charge. Other effects should be included, 
such as the role of the free electrons which might possibly exist, the 
effects of charge exchange collisions, etc. Moreover, experimental 
measurements of mobility are also difficult to perform. Because of 
these complexities, the most comprehensive calculations to date have 
not been fully borne out by experimental observations. 

Even though we can presumably manufacture particles which follow 
the gas because of their low mobility (i.e., no slip), we must somehow 
also arrange that the relative density of the flow of charged particles 



be sufficiently high. Obviously, if there are only a few charged 
particles, this will translate into a small amount of electrical energy. 

The theoretical relations governing power generation for an 
idealized case are given in Appendix A. These show the basic importance 
of mobility and relative density as mentioned above. 

Other fundamental problems pertain to particle losses and frictional 
losses. A judicious choice of velocities and channel geometries will 
be necessary in order to minimize these losses. It is interesting to 
note here the "slender" versus "broad" channel controversy. If we 
consider the flow in a generator channel to be axi symmetric, then we 
have to contend with a radial as well as an axial component of the 
electric field as obtained from Poisson's equation. In a "broad" 
channel the axial spacing is small so that one may neglect the radial 
motion of the ions. But this configuration has a pressure drop limited 
by the breakdown field to an unattractively small value. The "slender" 
channel attempts to increase the pressure drop by lengthening the 
channel and accounting for radial ion movement to the walls. Neither 
configuration has proven ideal but the problems here have received 
more attention than those mentioned in earlier paragraphs. 

C. Method of Approach 

Our approach has been to focus on the fluid dynamics of an EHD 
generator utilizing a gas for the carrier medium. The need to build 
an experimental facility, however, has prompted us to consider all of 
the relevant EHD effects. A review of the work of other investigators 
was initiated (see Chapter II) and a facility which operates with air 
at subsonic speeds has been constructed (see Chapter III). 

An important objective of our program is to study the influence of 
turbulence on the EHD process. Present evidence suggests that this 
factor has a very significant effect on the operation of EHD generators. 
A highly turbulent flow can be shown to yield favorable results in 
increasing the breakdown potential of the carrier gas. It also has the 
collateral effect of strongly affecting the mean effective mobility of 



the charged particles. The main disadvantage associated with turbulence 
is the larger fluid dynamic losses involved. Figure 2 illustrates 
schematically the application of this principle. The role of turbulence 
is discussed further in Chapter IV. 

Since the manner by which charges are introduced into the carrier 
gas is the most critical in the EHD process and the most complicated in 
practice, the injector is another main concern in this program. 
Analysis based strictly on mobility arguments can be used 2 to show a 
certain optimum range of charged particle sizes. The mobility relations 
in terms of frictional forces, as given in Appendix B, are relevant in 
this regard. Also, on the experimental side we have operated two types 
of injectors, namely, molecular and two-phase, and conclude that the 
two-phase injector has a good potential for efficient operation. 
Appendices C and D are two Master Theses done at the Naval Postgraduate 
School in this connection. 

Chapter VI presents some tentative conclusions from this work. 
Our proposed endeavors for the year 1969-70 are mentioned in 
Chapter VII. 



II Survey of Work in the Field 

The principles of EHD generation are based on phenomena which have 
been studied for over a century. Armstrong (1843) found that a steam 
jet which contained minute water drops would charge up an insulated 
body. He found no such effect with dry steam. Babat (1936) operated 
a generator using a supersaturated mercury jet which picked up charged 
particles as it passed through a discharge. Babat actually generated 
about 5 watts but since the particles moved in a collisionless regime, 
his was actually a "ballistic" generator. M. Mureau-Hanot^-* charged 
coal and glass dusts in order to obtain more massive charge carriers. 
Recently, W. E. Bennett (1959) revived the idea of the gaseous 
Van de Graaff with some experimental results and with a discussion of 

the advantages of such generators. 

25 
0. Stuetzer has been active in the field of EHD for many years. 

His early concerns have been the pumping of liquids (Ion Drag Pumps) 

and he has made contributions to the ion injector problems and to the 

understanding of space charge flow. More recently, he has done work on 

EHD flow control which consists of an electrically induced transition 

from turbulent to laminar flow and some control on boundary layer 

separation. All his work has been with insulating liquids. H. R. Velkoff 

of Ohio State University has done wome work on EHD effects on gas flows. 

The survey that follows is intended to be neither comprehensive nor 
exhaustive since the work has been summarised elsewhere. » » i:> » We 
intend to present here a partial account of generator work which is 
largely derived from impressions during visits to the various companies 
and laboratories presently doing EHD research. 

Pioneering efforts in EHD power generation were made by Alvin Marks 
who proposed the use of a charged aerosol. ^ He recognized the importance 
that ion mobility plays in the performance of EHD devices. Marks has 
worked on generators and smog control units and holds many patents. He 
has proposed a "mixture-condensation" process^ for the formation of 
the charged aerosol which attempts to overcome some of the difficulties 



26 



inherent in the generator. Marks forsees the EHD generator breaking the 
monopoly of large utility companies by making possible the generation of 
smaller amounts of electrical power economically. He has also proposed 
that a fluid with low mobility, unipolar ions be studied as a new 
thermodynamic working medium. 

The Air Force has sponsored several programs in this field with 
a strong in-house capability (WPAFB) headed by Dr. H. von Ohain. 
Contributions of his group include an electrostatic analysis of 
generator configuations which singles out regions of undesirably high 
electric fields and predicts the performance of a given configuration, 
the use of guard electrodes to improve injector operation, and the design 
and operation of high pressure hardware. The scaling laws for EFD 
(electrofluiddynamic) generators based on a simplified version of the 
physics have been formulated. A two-fluid system (ejector cycle) has 
been proposed to overcome the low pressure ratios of the generators. 

Dr. M. C. Gourdine in the late 1950's carried out a one dimensional 
gas dynamic analysis of EHD flows. ^ He concluded that the energy 
extraction was so minimal that the gas dynamic equations could be 
uncoupled from the electrodynamics. His analysis shows that the electric 
field can have its highest value at the injector and, thus, breakdown 
there may be a problem. Later he became a proponent of the "slender 
channel" generator configuration.^* 1 ^ Briefly, in an axially symmetric 
flow, the axial electric field (x-direction) is assumed relatively 
constant so that Poisson's equation can be simplified by an order of 
magnitude argument, i.e., 



H*"il?< rt r> < 2 "» 



Theoretical predictions made on this basis show an advantage over 
"broad" channels, but these predictions have not been borne out by 
experiments. A large scale generator system has been attempted using 



coal dust charged in a corona discharge. Louvers are used to "contain" 
several parallel slender generator channels in a connnon vessel. 
Selected charge-bleeds to ground from these louvers are used in an 
attempt to minimize the effects of charge build up. Gourdine has 
marketed an electrostatic paint spray-gun which has a built-in EHD 
generator which provides all the high voltage electricity needed for 
the operation of the gun. 

An allied but separate effort has been carried out at the Curtiss- 
Wright Corporation under several investigators. 7,13,14,15 They have 
worked with the slender channel with a charged air-water aerosol at 
both subsonic and supersonic speeds. Charged particle mobility 
measurements have been studied. Since the one-dimensional analysis of 
the generator has met with little success, a two-dimensional one has 
been carried out which accounts for the radial space charge field. 
Two-dimensional effects are shown to affect the generator load 
characteristics, although the theory agrees only qualitatively with 
experiements. 

None of the above efforts, however, have convincingly shown that 
a substantial amount of flow energy has been transformed into electricity, 
Many ingenious schemes have evolved from these efforts, but progress in 
solving basic problems has been sporadic. 



Ill Hardware 

We have chosen to use air as the carrier medium because of its 
accessibility in the laboratory. A Carrier three-stage centrifugal 
compressor has been used which has a maximum flow rate of A, 000 CFM 
(inlet air). Its maximum pressure ratio is two and the exit air 
temperature can be varied from 80° to 240°F with an after cooler. 
While free from pump oil, the air contains moisture and whatever 
impurities it has at the inlet which do not precipitate at the pump. 
Before entering the test apparatus, the air is metered with a sharp- 
edged orifice built to ASME standards, and its temperature is also 
measured. 

The test region is preceded by a plenum which serves to reduce the 
level of free stream turbulence to an intensity of 0.14% RMS in the 
test section. The flow straighteners in the plenum consist of a series 
of honeycomb and screen sections. In this plenum, additives to the flow 
may be introduced. The plenum is followed by a nozzle and then the test 
section. The test section is a rectangular channel 6" long of cross 
section 1" by 2". Typical runs are made at a Mach number of 0.3 and 
atmospheric conditions. The building material for the plenum and 
test section is Plexiglass, chosen for its excellent insulating 
properties, its ease of fabrication, and its transparency. This facility 
is described by Wallace (Appendix D) . A new plenum-test section has 
been constructed which has increased the test section dimensions to 
2" by 4". In addition, it permits the removal to the test section from 
the nozzle. This is shown in Figure 3. 

The turbulent region of interest is the wake of a cylinder. The 
injector unit is incorporated into the trailing edge of the cylinder. 
Of the two types of injector tried the one that supplies air ions, of 
course, is the simpler to manufacture. It consists of a needle imbedded 
in the cylinder and recessed from the ring so that the electric field 
produces a velocity which is partially in the flow direction. As would 
be expected, these ions are too mobile in spite of the turbulent field. 



10 



This type of injector is discussed in Appendix C. The two-phase 
injector consists of a corona discharge within a metallic nozzle where 
supersaturated steam is accelerated to sonic-velocities. The initial 
unit is also reported in Appendix C. The results obtained are a 
considerable improvement over the air ions, but they are not as good 
as we had hoped. Hence, a new nozzle was built of teflon with a metallic 
insert at the throat which acts as the corona ring. This causes a more 
stringent condition for the corona operation since the spacing between the 
ring and needle is quite influential as to whether or not breakdown occurs 
before the onset of the corona current. However, better success has been 
achieved with this unit and it is the subject of Chapter V. The 
generator mode investigated so far has been the "short-circuit" case 
because we are interested primarily in injector performance at this 
time. We may note here that the internal impedance of the generator is 
so high that a usual type of application will not noticeably load it 
(except for the condition of load matching for maximum power). The 
diode bank used in conjunction with the collector has a resistance of 
about 3 megohms (see Figure 8, Chapter V) which is low compared to the 
generator internal impedance. The diode bank calibration curve is 
shown in Figure A. 

The injector does not fill all of the channel cross section and 
leaves ample space between the conversion region and the channel walls, 
as depicted in Figure 5. The flowing air is an excellent insulator and 
no problems have been encountered with shorting. The collector is a 
traversable needle mounted through the back of the channel. This needle 
collector offers less resistance to the flow than a screen. It was 
found that the sharp needle would back-corona and hence overshadow the 
EHD process. Subsequently, a guard ring was unsuccessfully installed. 
This ring can presumably shield the corona high voltage (at the nozzle 
exit ring) from the collector which, like the corona needle, is grounded. 
It is conceivable that the geometry of the injector-collector-guard 
ring system could be improved, but since we wanted to traverse the 



11 



collector this seemed like a rather Involved operation. Instead, the 
series of diodes (or electronic valves) were installed which force the 
needle to operate in the generator node. More recently we have blunted 
the collector tip somewhat. 

Mappings of the collector current vs. space are not complete to 
date but we have observed small changes of planes cross wise to the 
flow and an expected decay in the axial direction. The collector area, 
in any event, is very difficult to define because of the back-corona 
effect, and because water condenses everywhere and particularly at the 
collector probe. 

The high voltage is supplied by a Sorensen High Voltage DC power 
supply, with a range of - 30 KV and - 20 ma and with adjustable 
scales on its meters. It has a relay that disconnects the supply when 
a current surge exceeds 2 ma. All high voltage leads are properly 
labelled and provisions for high voltage safety are incorporated in 
the room. The high voltage connections are terminated in spherical 
brass balls so as to minimize corona discharges. J All cable used is 
good in excess of 40,000 volts. 

Two high voltage voltmeters are available, a Singer (Sensitive 
Research) electrostatic voltmeter which measures - 40 KV in four 
scales and is calibrated to 17, of full scale, and a High Voltage 
Engineering generating voltmeter which has essentially infinite DC 
impedance. This generating voltmeter has been mounted in a set of 
Rogowski electrodes to insure a homogenous capacitance. 

Currents have been measured by a Calico digital multimeter, a 
Keithley Picoammeter, and an assortment of Simpson mlcroammeters. 
These current measurements have always been made to ground. 

Additional measurements of pressure, temperature, and humidity 
are made with conventional metering equipment. 

For the study of turbulence some additional special instrumentation 
has been obtained. A voltage divider and current probe, which 
measures DC to 50 MHZ, have been obtained for Tektronix oscilloscopes. 



12 



In addition a Hewlett-Packard X-Y display scope and a General Radio 
real tine analyzer have been purchased. 

High pressure Heise manometers for future expansion into above 
atmospheric operation are available and metallic and nonmetallic sub- 
micron powders have also been purchased for future use. 



13 



IV Effects of Turbulence 

The definition of the effects of fluid dynamic turbulence is a 
significant portion of this program. Bela Karlowitz , in his work 
on enhancing the enthalpy of flames by the addition of electrical 
energy, has found that strong, free stream turbulence acts to suppress 
filamentary discharges in plasmas and hence to increase the breakdown 
potential. We propose to study how turbulence affects breakdown and 
in particular, how it affects the behavior of free electrons, ions, 
and charged particles in the flow. There are always some free electrons 
present in a space charge flow, and because of their lower mass they can 
be accelerated to ionization energies much more readily than ions. An 
appropriate cumulative effect vould cause enough volume ionization to 
short-out the high voltages necessary for EHD operation. Breakdown 
of the carrier gas is a severe practical limitation and an increase 
of the breakdown potential of air by a factor of ten would expand the 
applications of EHD to areas such as boundary layer control. 

The analysis presented here is somewhat heuristic. Because of 
the high electric fields and the large disparity of mass of the 
charged species (electrons, ions, and particles) the definition of 
mean effective transport coefficients is somewhat difficult. Moreover, 
a rigorous statement of the problem would involve a coupling of the 
equations of turbulent flow with the electrodynamic equations. Needless 
to say, this is a complicated task, and we shall not undertake it here. 



Let us first look at the case for a plasma, where the diffusion of 
charges is ambipolar. The ratio of the diffusivity of momentum to 
that of mas is given by the Schmidt number (Sc) defined as 

S c ) £ - v/D (laminar) (4-1) 

S c ) t - (e y + v)/(e m + D) (turbulent) (4-2) 



14 



where v - molecular diffuaivity (momentum) 
t - eddy diffuaivity (momentum) 
D ■ mass or charge diffuaivity 
e m ■ eddy mass or charge diffuaivity 



27 
Now, according to the Reynolds analogy , 





v m 


and 


E >> V 
V 




e >> D 
m 


so that 


s c ) t = 1.0 



(4-3) 
(4-4) 
(4-5) 
(4-6) 



That is, the turbulent dif fusivities for mass and momentum are said to 
be the same at any point in the flow and much larger than their 
corresponding laminar values. The turbulent coefficients, unlike 
their laminar counterparts, depend strongly on the flow field. The 
laminar flow coefficients are functions of molecular phenomena and, 
thus, are true coefficients. Note that simple kinetic theory predicts 
the Schmidt number to be unity. This suggests that the simplified 
mechanisms for momentum and mass transfer are the same for laminar and 
turbulent flows. We expect the Reynolds analogy to be a useful tool 
even though it may not be entirely accurate in the present case. 

Let us consider air at standard temperature and pressure flowing 
around a cylinder, flte acknowledge the fact that this is a "poor" plasma) 

v - 1.5 x 10~ 5 m 2 /sec 

e - 0.016 \J m d (wake of the cylinder) 16 (4-7) 

and take U. - 100 m/sec (free stream velocity) 

d ■ lO""^ (cylinder diameter) 

so that e ■ 0.016 nr/sec * e 
v m 



15 



3 
Note that the turbulent diffusivity is larger by a factor of about 10 

than the molecular value. Under our present treatment, this fact is 
not expected to affect greatly the Schmidt number of the ions. However, 
the electrons because of their low mass do not exchange momentum by 
collision with neutrals very efficiently and hence are somewhat uncoupled 
from the flow pattern (this is the same argument used in the two- 
temperature plasma). In the other hand, particles are more highly 
coupled than ions because of their large size and we might expect a 
change of the Schmidt number. 

We will assume that the Einstein relation ' between the diffusion 
and the mobility applies to the charged species. 

D - t » " n^>6 » " «" 8 > 

T - temperature (°K) 
|j ■ mobility 

The above relation expresses diffusion as motion due to thermal energies. 
More accurately, we can state that the ration D/p represents the 
average energy of the species. For the ions and particles it is given 
by the gas temperature. For the electrons, however, it is considerably 
above the gas temperature because of the presence of the electric field. 
For molecular air ions at 300°K, the ion mobility is" 

-4 2 
p. ■ 2x10 m /volt-sec 

whereupon the Einstein relation gives 

D ■ 5.2 x 10 m /sec 

whereas for a 0.1 pm diameter particle we have (see Fig. 1) 

-7 2 
p ■ 1x10 m /volt-sec 
P 



16 



-9 2 
and D - 2.6x10 m /sec 

P 

We are picking this size particle merely because it is a representative 
one, somewhat independent of the method of charging, rather than 
because it is an optimum. 

For electrons, the situation is somewhat more complicated since 
the mobility is a function of the electric field. However, for an 
electric field of the order of the breakdown potential in air 
(3xl0 6 v/m) we can estimate the mobility to be 9 ' 10 



and 



2 
y * 0.1 m /volt-sec 
e 



-3 2 

D a 2.6 x 10 m /sec 

e 



(This is perhaps not fair in the sense that the breakdown potential of 
a plasma is actually lower than that for un-ionized air, but it will 
suffice for now.) The electron mobility at lower fields is higher than 
the value quoted. There is also some question as to whether or not 
the electron temperature should be higher than the gas temperature 
because of the electric field but, again, we will not dwell upon that 
here. 

The resulting Schmidt numbers are shown in Table I. 

TABLE I Schmidt Numbers for Ambipolar Diffusion 
Mobility (m 2 /v-sec) (S c > £ (S c > t 



"1 



A. 


IONS 




2 x 10" A 


B. 


PARTICLES 




1 x 10" 7 


C. 


ELECTRONS 




, « ,«-l 



5,800 *1(?) 



1.0x10 * 5.8xl0" 3 5.8 



17 



For the ions, Table I shows that there is no substantial difference 
between the laminar and the turbulent case. The apparent undesirable 
increase of diffusion in the turbulent case is not large enough to be 
meaningful under our stated approximations. Experience shows, moreover, 
that the drift velocity of ions in both laminar and turbulent fields is 
too high so that the ambipolar conditions must not fully prevail in our 
case, i.e., we would anticipate that the Schmidt number of ions would be 
less than one. 

The charged particles, as expected, appear quite strongly coupled 
to the laminar flow field (S c >> 1) but seem somewhat uncoupled in the 
turbulent case. This, of course, is a direct result of Reynold's 
analogy. Experimentally, particles perform rather well in both laminar 
and turbulent flows which casts some doubts on the validity of the 
result shown in Table I. One may argue, for example, that ambipolar 
conditions do not prevail. There is also another point to be considered, 

namely, the equilibration time for the particles. As shown in 

20 
Appendix B, the e- folding time for a micron-size water droplet can be 

calculated to be 10 sec; turbulent fluctuations of frequency higher 

than 100 kHz could thus not be followed by the particles and it is 

doubtful that the diffusion coefficient could rise to the full value 

of the momentum coefficient. Certainly, particles bigger than a 

micron have much larger equilibration times. 

For electrons, the Schmidt number appears to be quite revealing. 

The laminar case shows that the diffusion rates dominate the momentum 

transfer rates (S c << 1). These electrons, hence, are not influenced 

by the flow field. The turbulent case shows that the opposite begins 

to be true and we see that the flow field can carry the diffusing 

electrons. In particular the flow field can alter the electron 

diffusion pattern in space and thus break up filamentary discharges. 

We have to note here that we have assumed no enhanced electron diffusion 

by the turbulent field because of the weak collisional coupling. 



18 



Let's look now at the case of space charge flow. The charged 
species equation (electrons, ions, particles) is 



r - -DVn + nuE (4-9) 

where T ■ specie flux 

n ■ specie density 

E « space charge field 

Using the Einstein relation 

T ■ »« «- T + iff V (4 - 10 > 

In the case of ions and particles, we can argue that the flux 
will be predominantly due to the space charge field, except for a small 
region in the neighborhood of the injector where the density gradient 
may approximate a step function (i.e., Vn ■ •») . This space charge 
predominance can be shown to exist for a Gaussian distribution of 
particles at the densities of interest. 

Now, for the electrons the electric field is the same; however, 
the number density is considerably smaller and the temperature (energy) 
considerably larger than that for the others. This can be argued to 
mean that, for electrons, diffusion can be important over a considerably 
greater region of space than for the ions or particles. Even if this 
region does not cover the entire typical dimension of the EHD channel, 
it will suffice to assume that this region is finite in extent and, 
hence, we may say that there is some region where our results for 
ambipolar diffusion apply. Electron densities can be quite low in a 
space charge flow because of the absence of sources at the surfaces 
and because of high recombination rates in the gas. 

Since we assume that the space charge diffusion is dominant for 
ions and particles, let's estimate the Schmidt number on the basis. 



19 



We can say, 



and 



nuE s - wEgt (n/*) (4-11) 

tB = uE c £ ("diffusion" coefficient) (4-12) 



I ■ characteristic length 

If we pick t to be the apparatus length (1 cm) and E g to be 0.1E fe 
(3xl0 5 v/m) or pick I as 1 mm and E_ as 3x10 v/m, then 

<$ z 3xl0 +3 y 
which for ions becomes 

£ z 0.6 m 2 /sec 
and particles 

fo z 3x10 m /sec 

The results are shown in Table II. Note that we are somewhat 
arbitrarily carrying over the previous results for the electrons. 

TABLE II Schmidt Numbers for Space Charge Flow 
Mobility (m /v-sec) S ) S ) 



2.6xl0" 2 



A. 


IONS 








-4 
2 x 10 




2.5xl0 -5 


B. 


PARTICLES 








1x10" 7 




5xl0" 2 


C. 


ELECTRONS 


(Ambipolar) 






0.1 




5.8xl0" 3 



0.98 



5.8 



In the case of ions in turbulent flow, the Reynolds analogy seems 



to break down since the value of e is not much greater than 

m 



J. 



20 



We could, however, not argue further because of the arbitrariness of 
-ft at the present stage. 

The picture presented by Table II is quite different from Table I 
Both ions and particles see an improvement with the turbulent field 
and the particles are always more closely coupled to the flow than the 
ions. This should satisfy our intuition but, obviously, the ideas 
presented above are not complete and we must await the results of 
better calculations before feeling satisfied we understand the role 
of turbulence. 

We have consistently observed an effect of the flow with our 
air corona discharges. As mentioned earlier, the molecular ion 
injector was mounted on the wake of the cylinder, and a typical 
improvement of 200 to 400 volts in the breakdown potential was 
recorded. Breakdown, moreover, was arrested so that it generally 
never surged to the 2 ma needed to operate the trip-mechanism in 
the power supply. In the case of the steam injector, to be discussed 
in Chapter V, no corona operation could be obtained without the flow 
of steam in the nozzle, so that here turbulence (of a level not 
defined here) also plays a desirable role. Table III shows some of 
the increases in breakdown potential that we have measured as a result 
of the turbulence at the wake of a cylinder 1 cm diameter (see 
Appendix D) . There is some uncertainty in the role that impurities 
might have played in data reproducibility and, hence, we are 
refraining from making a more quantitative estimate of the role of 
turbulence here. The data are shown for various corona units. 



21 



TABLE III 
Effect of Turbulent Flow on Breakdown of Air Corona 



Type 
of ion 


No Flow 
Breakdown Potential 
- Kilovolts - 


Negative 


5.05 


Positive 
ii 


6.60 

it 


ii 


n 


Negative 


5.6 



Breakdown Potential (Free stream velocity) 
- Kilovolts (m/sec) - 



> 5.5 (100) 

6.8 (104) 

7.2 (72) 

7.0 (100) 

> 6.0 (100) 



22 



V Injector Work 

As was mentioned earlier, the injector is the most critical 
component of the EHD system. This is because the size, charge, and 
density of ions must lie within a narrov spread of the optimum values. 
Moreover, corona type injectors utilize electrical energy which must 
not be wasted in creating an assortment of ion sizes (even though we 
could provide a scheme to select out the correct ion to be Injected 
into the flow). We are pursuing the injector problem both analytically 

and experimentally. In our analysis, the definition of injector 

7 14 
parameters parallels that of other investigators. * 

The experimental challenge is perhaps the more crucial here. 
Some work (but not enough) has been done elsewhere* *^ in growing 
aerosols by condensation and in developing techniques for measuring 
ion size. We have so far concerned ourselves with the air corona in 
the turbulent wake and with the supersaturated steam injector. Both 
of these are described in some detail in Appendix C. In this chapter, 
we will be only concerned with the work on the steam injector. This 
injector is an improvement over the one reported by Ober. Supersaturated 
steam passes through a corona discharge which is made up of a needle 
inserted through the front of a Teflon cylinder and a ring imbeded at 
the exit of the nozzle. The spacing between the needle and ring 
can be adjusted. The nozzle is at the wake of the cylinder facing 
downstream and the cylinder provides a reservoir of steam for the 
expansion. In this new design the cylinder is made also of Teflon. 
The design is shown in Figure 6. Steam is fed from a boiler, through 
a heated line, to this cylindrical reservoir which at the same time 
is used to generate the strong turbulence region. 

This nozzle has been run with some success and Figure 7 shows data 
obtained by traversing the collector needle back and forth along the 
centerline of the nozzle. Here we plot the ratio of the current at 

the collector to the current issuing from the corona needle versus 

1 ft 
the corona voltage. Below about 1.4 KV the system gets into a very 



23 



unstable mode and above about 2.5 KV the field Is so strong that only 
a few percent of the total current is convected by the flow. This flow, 
incidentally, is about 1.5x10 J lbm/sec of steam and the conditions at 
the throat are sonic. The electrical circuit used to obtain the data 
of Figure 7 is shown in Figure 8. As mentioned previously, we had 
to use a diode bank to prevent the back-corona. If this unstable mode 
were not present the efficiency of the injector could be improved by 
lowering the corona voltage. This is certainly obvious from Figure 7 
for the smallest spacing (L » 0.075 cm). The higher spacings along 
the centerline show a maximum which is to be expected in such cases. 
In fact, Figure 9, shows a plot of these maxima versus distance and 

we may attribute the shape of the curve to the ion depletion in the 

1 ft 
free jet. All of these data were obtained with the main flow off. 

We have recently improved the injector performance by running 
it just above the condition where droplets form at the throat and 
short the corona. This condition might be referred to as maximum 
wetness inside the nozzle. We have also recently purchased a steam 
generator which permits continuous running, in contrast to the one 
hour limit of the previous unit. 

More work needs to be done regarding the condensation (see Hill^ 
and the Curtiss-Wright report 1 ^) of steam around the ionic nuclei 
created in the corona discharge, specifically towards a uniform 
formation of charged particles. 



24 



VI Summary, Conclusions, and Recommendations 

A. Summary 

Electrohydro dynamic 6 embraces a series of specialized fields of 
endeavor. A partial list of these is given below: 

1) High voltage technology — the large voltages involved 
require the best insulation available. But, beyond that, surface 
impurities such as moisture and corona discharges from sharp points 
can leak enough current to mask those currents being meaaured. High 
voltage, low current metering equipment is rather specialized and 
safety procedures are more involved than with usual electrical equipment. 

2) Two phase flow — the gas dynamics of two phase flow, 

20 
including condensing vapors (see Marble ) , is a relatively sophisticated 

field. The extension is needed to a charged aerosol flow. 

3) Injector — the injector has to meet some fairly 
exacting requirements as to ion size and charge. To do this job 
efficiently may be too demanding of some of the present schemes. 

4) Breakdown — air is the most convenient carrier medium 
for open EHD systems but its characteristics are marginal because of 
the low breakdown field. The presence of free electrons in the flow 
affect the breakdown. Electron sources such as sharp corners also 
affect breakdown. 

B. Conclusions and Recommendations 

Based on our work to date, the following conclusions can be put 
forth: 

1) Strong, free stream turbulence appears to increase the 
breakdown potential of the carrier medium, mainly through its effect 
on the free electrons. Our present estimates are somewhat intuitive 
and point out the need for (further) study of the fundamental relations 
involved. 

2) Air with electronegative additives and high turbulence 
may be an acceptable medium for EHD schemes. 



25 



3) The definition of optima for particle sizes awaits 
experimental confirmation. It is not necessary to begin with an 
efficient injector, but rather with one that manufactures the desired 
size and charge. This could be done by "filtering out" the undesirable 
ions and only injecting the proper ones. 

4) A confirmation is needed of the assumptions used to carry 
out EHD generator analyses. This will require some rather sophisticated 
diagnostics. 



26 



VII New Research 

The project is continuing through Fiscal Year 69-70 in basically 
the sane form it was initiated. The injector work has continued and 
a study of turbulent spectrum both with an EHD probe and a hot wire 
anemometer has been Initiated. The use of a Langmuir probe is 
presently under study in conjunction with Professor Woehler of our 
Physics Department. An investigation of the possible use of EHD for 
boundary layer control is being conducted. In addition, a theoretical 
study of the effects of the turbulence is being undertaken by Professor 
Gawain of our Aeronautics Department. 

Our aim is to better define some of the work that, by necessity, 
has been sketchily presented here. In particular, it is desirable to 
develop appropriate mathematical models to describe at least the 
major phenomena encountered in our system. 



27 




'O 






■q 



■o e 'Q 

(D9S-a^u) — Al!|iq<W 



fc 



28 



p 

<LU 

52. 



FT 

\i 

\i 
\ 
\ 




i\ 
l\ 
l\ 

l\ 
l\ 



IS 

iL 





1 










1 



LU 
Z 
LU 
O 

O 

5 



LU 

_l 
D 
CD 
QC 

3 



< 
LU 
QC 
I- 
</) 

LU 
LU 



x 



29 




u 

c 



CEZ 

QQ 

II 



CM 


<r> 


*- 


& 



fed 

uj % 



Q. 



CO 
Ul 



UJ 



ro 

UJ 



30 




cv i_(snoA)A 



\ 

\ 
\ 
\ 
\ 
\ 

\ 

\ 
\ 

\ 

\ 

Ob! \ 

3B\ 



LJ u. _ 
Q 



£ \ 




O 



UJ 

X 

o 
</) 

<r 

i 

LU 

Z 
UJ 

o 

Q 

I 
UJ 

lO 

UJ 



t 



32 




in 

v* — 

c «> 
o</) 




o 

LU 



LU 
< 

a: 

i 
< 

ID 

LU 

a: 



33 



• 16-r 



.14— 



.12-- 



10-- 



.08-- 



C9 



06-- 



.04-- 



.02— 



O L= 075cm 
A L = 0.50 cm 
□ L = 0.25 cm 
O L = 0.075 cm 




Defined in figure 8 



0.5 



O 
O 



o 

o 




A 



o 



°cf^ 



o 



o 

A 



O 



1.0 
V C (KV) 



A 



O 
A 



A° 



o 

o O 
oaO 



o 



O 
O 



% 



O A 



20 



2.5 



FIGURE 7 TEFLON NOZZLE STEAM INJECTOR 

34 



<*> 




O Q 






>- 
Q 

h- 
(/) 

or 
o 

h- 
o 
w 



O 



3 
O 
QC 

O 

_l 
< 

o 

q: 

i- 
o 

UJ 

_l 

UJ 

GO 
UJ 

q: 
u. 



35 



20-r 



.15-" 



K .10 — 



o 



.05"- 



025 



O 



+ 



O 



+ 



0.50 0.75 

Ucm)— •► 



O 



.00 



25 



FIGURE 9 MAXIMUM RATIO VERSUS SPACING 



36 



REFERENCES 

1) Marks, A. M. , "Economic Implications of Charged Aerosol and Dipole 
Technology," paper presented before the Senate Subcommittee on 
Antitrust and Monopoly, Sen. P. A. Hart, chairman (October 1967). 

2) Marks, A. M. , Barreto, E. , and Chu, C. K., AIAA Journal, 2, 
45 (1964). 

3) Gourdine, M. C, "Power Generation by Means of the Electric Wind," 
JPL TN 32-6 (1960); "Electrogasdynamic Channel Flow," JPL 

TN 34-5 (1960). 

4) Collier, E. L., and Gourdine, M. C, "Resistive EGD Channel," 
AIAA Journal, 6, 2278 (1968). 

5) Lawson, M. , and Wattendorf, F. (Editors) "Selected Topics in 
Electrofluid Dynamic Energy Conversion," AGARDograph 122 
(December 1968). 

6) Decaire, J. A., and Lawson, M. 0., "Electrofluiddynamic Power 
Generation Trends and Expectations," Eight Syrap. on MHD, 
Stanford, (1967). 

7) Kahn, B. "A Continuation of the Basic Study of Slender Channel 
Electrogasdynamics," (Curtiss-Wright Corp.) ARL 65-4 (January 1965) 

8) Alston, L. L. (Editor), High Voltage Technology , Oxford University 
Press (1968). 

9) Cobine, J. D. , Gaseous Conductors , Dover (1958). 

10) Loeb, L. B., Basic Processes of Gaseous Electronics , University of 
California Press, (1960). 



37 



11) Karlowitz, B. , "Flames Augmented by Electrical Power," Pure and 
Applied Chemistry 1, 557 (1962). 

12) Marks, A. M. , Patent No 2,638,555 (May 1953). 

13) Kahn, B. , and Gourdine, M. C, "A Basic Study of Slender Channel 
Electrogasdynamics," (Curtiss-Wright Corp), ARL 63-205 (Nov. 1963) 

14) Brandmaier, H. E., Dimmock, T. H., and Kahn, B., "Research on 
Power Generation Electrogasdynamic Energy Conversion," (Curtiss- 
Wright Corp), ARL 67-0008 (January 1967). 

15) Brandmaier, H. E. , and Dimmock, T. H., "Factors Influencing 
Electro-Fluid Dynamic Power Generation," Journ. Spacecraft and 
Rockets, 4, 961 (1967). 

16) Hinze, J. 0. Turbulence , McGraw-Hill (1959). 

17) Brown, S. C, Introduction to Gas Discharges , John-Wiley (1966). 

18) Witby, K. T., "Generator for Producing High Concentrations of 
Small Ions," Review of Scientific Instruments, 32, 1351 (1961). 

19) Hill, P. G. "Condensation of Water Droplets During Supersonic 
Expansion Nozzles," Journ. of Fluid Mechanics, 25-3 , 593 (1966). 

20) Marble, F. E. , "Gasdynamics of Condensing Vapors," (Calif. Inst, 
of Tech), ARL 69-0040 (March 1969). 

21) Megaw, W. J., and Wells, A. C, Nature, 219, 259 (July 20, 1968). 

22) Fuchs, N. A. and Stechkina, M. Trans. Faraday Sc. 58, 1949 (1962). 

23) Mureau-Hanot, M. , "Transport of Ions by a Gaseous Current," 
C. R. Academic Sciences, Paris (1939). 



38 



24) Bennet, N. E., "The Generation of Direct Current at High Potential," 
Research. App. in Industry, 12, 455 (1959). 

25) Stuetzer , 0. M. , Journ. App. Phys., 984 (July 1959); Journ. App. 
Phys., 136 (Jan 1960); Rev. Sci. Instr. 16 (Jan 1961). 

26) Velkoff, H. R., "Electrostatically - Induced Secondary Flows in a 
Channel," Int. Symp. on EHD, MIT (1969). 

27) Kays, W. M. , Convective Heat and Mass Transfer , McGraw-Hill (1966). 

28) Schlichting, H. , Boundary Layer Theory , McGraw-Hill, Sixth Ed., (1968) 



39 



APPENDIX A 

Idealized Analysis of Basic 
Energy Conversion Process 
in EHD Flows 

In this appendix we will discuss some of the relations pertinent 
to power and efficiency of an EHD generator. Here, we will also 
focus on the power loss due to Ion slip. 

1) Assumptions 

The carrier fluid moving at velocity Uc entrains charged particles 
and moves them at velocity Tip against the resistance E of the electrical 
field. Velocities are maintained constant by a driving pressure 
gradient G in the direction of flow. We consider particle sizes to 
be much larger than the mean free paths of the molecules in the carrier 
gas so that we can speak of a pressure force on the particle. 

The particles are treated as being all of identical volume 
6v , identical charge q and moving with identical velocity u£. 

Quantities E, G, U^, "up, A, are constants. 

The only loss considered in this analysis is that associated 
with the mobility or slip of the charged particles relative to the 
carrier fluid. Figure Al depicts the system under study. 

Steady, incompressible flow is assumed. Friction and viscosity 
are neglected except for the viscous drag of the charged particles 
moving in the carrier fluid. It is assumed that Stokes' drag law for 
a sphere applies. In the range of particle size considered, the 
spherical shape turns out to be a good approximation. 

2) Forces on a Particle 



F - Eq - Electrical Force (A-l) 



F„ ■ G6v ■ Pressure Force (A-2) 

G p 



A-l 



X 



\ 

\ 

_\ 

\ 

\ 
\ 
\ 
V 

\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 
\ 



J 



o 



3 



I if 



1 




< 

z 
< 

2 



LU 



UJ 



A- 2 



F 



- K <U„ - lip) - Viscous Force (A-3) 



where K ■ 6*11^ according to Stokes' drag law. 
Hence the equation of notion becomes - 

-Eq + G6v p + K (U„ - u p ) - <5™ p (-^) (A-4) 

3) Momentum Equation for Two Phase Flow 

We apply the momentum theorem to the working section of the 
duct, which lies between Inlet and outlet cross-sections. Fluid and 
particle velocities both remain constant, hence there Is no change In 
momentum. Viscous Interaction forces between carrier fluid and 
particles are equal and opposite on fluid and particles and vanish In 
the summation over both phases. Therefore 



(A-5) 



Pressure Electrical 
Force Force 



Hence 



n q E (A-6) Balance of pressure 

and electrical forces 



4) Mobility vs Viscous Drag Factor 

Mobility y is defined by the expression, 

(U„ - "up) E W E (A-7) 

subject to the restriction that U. and u_ have reached constant limiting 
values. Under these conditions (A-4) becomes 

- Eq + G6v~ + K (U„ - u ) - (A-8) 

~» P P 



A-3 



Now substituting (A-6) and (A-7) into (A-8) gives 



- Eq + n q E 6 v + K u E - (A-9) 

P P 

therefore 

v - | (1 - n 6v) (A-10) 

K p p 

which defines the relation between u and K. 

Consider the limiting case where 6m_6v -* then 

u - y ft - | (A-ll) 

O K. 

substituting this back into (A-10) gives 



u - u (l-n«v) (A-12) 

o p p 

where y is a constant. This shows that apparent mobility p is a 

function of the volume ratio n 6v ! 
P__P_ 

5) Power Relations 

The ideal useful electrical power delivered by the charged 
particles moving against the resistance of the electrical field is 



P e - n At q E u (A-13) 

The ideal fluid power expended in pumping both the carrier fluid 
and the particles is 

P f - Gl [Q p + Q f ] (A-14) 



A-A 



where Q - n 6v A u Volumetric flow rate (A-15) 

p p p p of particles 

Q f - (1 - n 6v ) A U. Volumetric flow rate (A-16) 

p p of carrier fluid 

Substituting (A-6) , (A-15) and (A-16) into (A-14) and rearranging 
gives 



P, - n A I q E 
f P P 



I U. - n p 6v p (U. - u p ) C (A-17) 



Now substituting (A-7) into (A-13) and (A-17) gives the two 
expressions — 



P 

AT " p p 



- n q e U. I 1 - J?' 1 r.. a ^ (A-18) 

p M " \ U.,, J Electrical 



power density 
At * P p ' 



v EU - ^-%^{€)} 



Fluid power density 

The ratio of these two quantities fixes the efficiency of this 
idealized energy conversion process, that is 

P 1 -l^) 

rV . , r-.tjrES (A " 20) 

f / 1 - (n p 6v p ) (-] 

Moreover, we may substitute (A-12) into (A-18) and (A- 20) . The 
results are — 



(_ — £__ )- P * . i - (1 - n 6V ) [-£- 
Aln qEU„ / e p p ' V U. ) 



(A-21) 



A-5 



and 



V n E 



V 



For the ideal case of zero mobility, 



* 

P -► 1 

e 



Vi 



1 - n «V (1 - n «▼ ) -17- 

P P P P V u- 







o and 



(A-22) 



(A-23) 



6) Relative Power Loss Due to Slip 

P. 



Let 



e Atn q EU W 



(A-24) 



f Aln q" EU. 
P 



(A-25) 



The relative slip loss may be defined as 



A* 



* * 
p * - p i 

f e 1 , 

T - 1 



(A-27) 



Substituting from (6-10) , we obtain — 

2 K E 
I 



"-W 2 1-fc 



(A-28) 



A-6 



u E 
Eqs (A-21) and (A-28) show Che importance of minimizing o /U„ 

and maximizing n 6v . It can be shown that for the range of micron 

^ u E 

size particles with between 1 and 100 charges the ratio o /U- is 

acceptably small. Moreover, for n of the order of 10 xv m , the 

product n Av is negligible so that micron size particles meet the 

desirable criteria defined above. A realistic estimate of n can be 

P 
made in terms of limitations at the injector (see Appendix B) . 



A-7 



APPENDIX B 

Charged Particle Mobility and Diffusion 

I. We will concern ourselves with particles in the micron and 
submicron diameter range for which the flow is in the low Reynolds 
number regime (creeping flows). Classical calculations (such as the 
Stokes drag law^° for a sphere), based upon the assumption of a 
continuum fluid with no fluid slip at the particle surface, are not 

valid as the particle radius approaches the order to magnitude of the 

21 
mean free path of the gas in which it is moving. Megaw and Wells 

have carried out mobility measurements with singly charged spherical 

polystyrene particles, which they compared to a theory due to Fuchs 

22 
and Stechkina , as shown in the Figure Bl. The theoretical line lies 

slightly above most of the experimental points. 

As a comparison, the prediction based upon Stokes 1 drag law for 
a sphere is also shown on the figure. The data are seen to lie 
generally between the two theoretical curves. A possible explanation 
is that the Fuchs and Stechkina theory allows for too much slip at the 
particle surface, giving a high value of mobility; whereas the Stokes 
theory, which assumes no slip, gives a low value for the mobility. 
Note that the difference between the two theories for a 0.1-mlcron 
particle is a factor of two. 

The calculation of the mobility of a charged sphere in air 
based upon Stokes' drag law is as follows: The force acting upon a 
particle carrying "q* charge in an electric field E is 



F_» q E (newtons) 

(B-l) 

q ■ Z e (coulombs) 



Z - number of electronic charges 
e - charge of an electron 



B-l 



Experiments of Megow ond Wells (Polystyrene) 

Theory of Fuchs and Stechklno 

Stokes Drag Law For a Sphere 




0.4 0.6 0.8 

Particle Radius ^m) — 



1.0 



1.2 



FIGURE BI.ELECTRICAL MOBILITY OF SINGLY CHARGED PARTICLES 

IN AIR AT STANDARD CONDITIONS 



B-2 



The Stokes' dreg Is 



where 



end 



D - 6* Ry„ (U. - u ) 

"■*• p' 

H • particle velocity 
U. » gas velocity 
R - particle radius 

V m " g*« viscosity 



(newt on ») 



(B-2) 



Mobility \i is defined as 

v„ - u 



W 2 



(B-3) 



In steady motion, which is assumed in the definition of mobility 



V 



so 



6ic Ru„ 



(B-4) 



For air at 20°C and 760mm Hg 



then 



u„ - 1.83xl(T Kg/m sec 



i_ 2 
M - 2. 94 x 10 q/R m /volt sec 



where q is in coulombs and R is in m. 

For a particle charged with one electronic charge 



4.7 x 10 



-10 



m 



volt-sec 



(B-5) 



where R is in microns. Equation 5 is the basis of the theoretical solid 
curve shown on the graph. This curve compares favorably with Figure 1 
of the text. 



B-3 



For design purposes, either the simple Stokes' theory or the 
theory of Fuchs and Stechkina may be used. The actual mobility can 
reasonably be expected to fall between these two values. A more 
accurate evaluation of mobility, expecially for particles whose radius 
is less than about 0.1 microns, must await additional experimental data. 

II. The distance that a particle must travel after being injected 
into the flow can be calculated with the aid of a few assumptions. Let 
us say that the ions are injected with no velocity component in the 
flow direction and that the Stokes drag law applies to this particle 
(since, whether or not the main flow is turbulent, the Reynolds number 
based on particle diameter is low). Then, equating forces, 

du 
6 w u„ R (U. - u p ) - 6-i p u p ^ (B-6) 

x ■ flow direction 

6m - mass of particle 
P 

The characteristic time scale for which the droplet reaches the 
gas velocity or e-folding time 20 is 

6m 
T - E_ 



v bTTRy. 
So that for x ■ 0, u ■ we have 



ln i - yu. ; U P /U - ■ £r1§ p 



(B-7) 



but 



P.v 



(B-8) 



% " P p ( 3 * r3) <M) 

P " particle density 



B-4 



Hence 



where 



6*P«>R x . 1_ _° * (B-10) 

P e p 

R = u » < 2R) (based on the diameter) (B-ll) 



Then, we can solve for x/R as 



k {* {.t^nz) - V"-} (B " ] 



Taking u /U,, - 0.99 since the free stream velocity is approached 
asymptotically, we have 

^ 0.402 ^ R e for ^ - 0.99 (B-13) 

For water droplets in air and a Reynolds number of 1.0 we get 



■jp - 328 (B-U) 

So that for a micron particle, x* will be 0.16 mm which is a 
relatively small distance. For higher Reynolds numbers (higher velocities 
for the same particle size) , the distance x* will be proportionately 
higher but the Stokes law will begin to break down. Still, it is 
noteworthy that for channels of the size of a few millimeters, the 
distance over which micron size particles reach the gas velocity may 
not be negligible. 

III. Let us treat briefly the space charge expansion due to its 
own electric field as treated in Chapter IV. We assume that 

V D " - ^T + uE S ' uE S <B - 15) 

where v_ ■ drift velocity (ions or particles) 



B-5 



Nov let's look at a spherical cloud of space charge. If the total 
charge Is Q (In coulombs) and there are many ions or particles, then 
a surface ion can be taken as a test charge and the electrical force on 
it will be 

F - ^ W " |> - q 8 (B-16) 

4 n e r Airer 
o o 

r - distance to the center of charge (radius of sphere), hence 



E r " 2 (B " 17) 

A ir e r 
o 

where E is the radial field on the surface ion due to the space charge, 
The expansion of the charged cloud will be radial, so that 



I'",' I 
Integrating we get 

o 
where r is the radius at t » 0. Rearranging, 



(r/r ) 3 ,1+4 -M UA t 



•' " 1 + * t 71 ' (B - 20 > 



o 



f 



o 



1 + £- (qn ) t (B-21) 



£ o P 

(qn ) - initial space charge density (corresponding to r ) 
P o 

Now for a constant velocity (U.) of the space charge cloud in 
the axial direction 

U„ - ~ x-0att-0 (B-22) 

«'V 3 " * * t «V„ t (B-23, 



B-6 



Introducing the gas density 

«5» - nu n„ (B-24) 

we get 

(r/0 S -i+J5*»X.(i\ i. (B-25) 



o e Q m. vja. Q U« 

We can find by the equation above the spread of the space charge cloud 

as a function of x given p, n /n_, and U_. Note that the product u6«e 

P 
is pressure independent. The expansion, then, will be as the ratio of 

radii cubed. 

In order to prevent breakdown of the carrier gas, we must demand that 

(qn p ) (j*r o 3 ) 

2 < E b (B " 26) 

4 i e r 
o o 

where EL is the breakdown field of the carrier gas 

E b /n - 

or r < 7= — r— 7 (B-27) 

o (qh ) /n-e 

P ° 
Using q~ ■ Ze we can evaluate the expression above for air. 

r o> • zlf/O lmtm) (B ' 28) 

max p ^° 

E, /ru» is relatively constant between one and 100 atmospheres. The 

relationship for r ) is representative of a typical injector size and 

says that any initial radius of the space charge cloud greater than r n ) 

max 

will cause a breakdown. 



B-7 



APPENDIX C 



United States 
Naval Postgraduate School 




THESIS 



ION INJECTORS FOR SINGLE- AND TWO- 

PHASE ELECTROGASDYNAMIC GENERATORS 

by 

William Taylor Ober II 



June 1969 



TluA document ka6 been approved &o\ pubtic n.z~ 
iea&e. and 6 ate.; itb duViibution -U wlimite.d. 



pages CI - C50 



Ion Iniectors for Single- and Two- 
Phase Electrogasdynamic Generators 



by 

William Taylor Ober II 

Lieutenant ( }unior grade), United States Navy 

B. S., United States Naval Academy, 1968 



Submitted in partial fulfillment of the 
requirements for the degree of 



MASTER OF SCIENCE IN AERONAUTICAL ENGINEERING 

from the 

NAVAL POSTGRADUATE SCHOOL 
June 1969 



ABSTRACT 

Systems suitable for the injection of ions into electrogasdynamic 
(EGD) generator devices were built and tested. The mechanism of injec- 
tion was based on a corona discharge, whereby ions moving through an 
electric field can be intercepted by a gaseous flow. The intercepted ions 
are of one polarity, insuring selective ion injection. Two types of in- 
jector units were investigated. One was a molecular ion device which 
produced ions directly from the carrier gas, and the other created 
larger sized ions, resulting in an aerosol flow. The latter consisted 
of passing saturated steam through a corona discharge and injecting it 
into an air stream. In order to aid the injection process, the wake of 
a cylinder in the air stream was utilized in both cases. Most of the 
work done here was devoted to the design and testing of the aerosol flow 
device. The degree of success was moderate. 



TABLE OF CONTENTS 

I. INTRODUCTION 9 

II. MOLECULAR ION EXPERIMENTS 11 

III. COLLOIDAL ION SOURCE 13 

A. INTRODUCTION 13 

B. DESIGN OF COLLOIDAL ION GENERATOR 15 

C. EXPERIMENTAL PROCEDURE 16 

IV. ANALYSIS OF RESULTS 19 

V. CONCLUSIONS 22 

VI. RECOMMENDATIONS 23 

APPENDIX A CALCULATION OF NECESSARY STEAM FLOW RATE 42 

APPENDIX B CALCULATION OF NOZZLE EXIT AREAS 43 

BIBLIOGRAPHY 46 

INITIAL DISTRIBUTION LIST 47 

FORM DD 1473 49 



LIST OF ILLUSTRATIONS 

1. Generator schematic 24 

2. Preliminary corona device 25 

3. Molecular ion device 26 

4. Corona current vs. voltage (molecular ion unit) 27 

5. Electrical setup of molecular ion injector 28 

6. Corona current vs. voltage (molecular ion unit) 29 

7. Colloidal ion generator . 30 

8. Steam generator schematic 31 

9. Photographs of steam unit and pressure cooker 32 

10. Corona current vs. voltage (micron injector unit) 33 

11. Corona current vs. voltage (micron injector unit) 34 

12. Electrical schematic of micron injector unit 35 

13. Photograph of generator in test section ..... 36 

14. Photographs of multimeter and diode series 37 

15. Generator current vs. distance 38 

16. Generator current vs. corona current 39 

17. Generator current vs. corona voltage 40 

18. Epoxy coating on nozzle 41 

19. Nozzle dimensions 45 



LIST OF SYMBOLS AND ABBREVIATIONS 

E, electric field strength 

EGD electrogaadynamic 

I corona current 

I generator current 

I-V current -voltage 

L length 

V corona voltage 

7/- charged particle drift velocity 

yU. charged particle mobility 

W mass flow rate 

A* nozzle exit area for choked flow 

k C /C 

P v 

R steam gas constant 

P pressure 

o r 

T temperature 



ACKNOWLEDGEMENT 

The author wishes to express his deepest gratitude to Professor 
Oscar Biblarz for his invaluable guidance, to Pat Hickey for all the 
time and effort he devoted to this project and to Lieutenant (junior 
grade) Dave Wallace for his aid in co-ordinating this effort with his 
own. 



I. INTRODUCTION 

The principle of operation of an electrogasdynamic (EGD) generator 
is based on the fact that if a moving gas can displace ions away from 
their point of creation it can perform electrical work. The generator, 
therefore, consists of an ion injector, a conversion region (where the 
ion migration path is influenced by the direction of the gas flow), and 
a collector device to pick up the ions which have been carried down- 
stream. Of these three basic components, the ion injector is the most 
complex and critical to the proper operation of the generator. The 
purpose of this work was to design and test two means of ion injection, 
classified by the nature of the ions desired in each case. The first 
injects molecular air ions into an air stream. The second one injects 
ionized water droplets into the air stream; this injection method will 
be alternately referred to as either a colloidal ion suspension or an 
aerosol flow. 

The problem of creating an efficient EGD generator has been studied 
with considerable interest over the past seven years • It was decided 
at the start of this project to employ a cylinder mounted spanwise in a 
subsonic test section to generate a two dimensional turbulent wake. 
It was thought that this wake would facilitate the removal of charged 
particles from their generation point and thus aid the process by which 
the air flow forces them downstream to a collector device. With this 
goal in mind all designs for the generation and injection of charged 
particles were incorporated with the cylinder. It was also felt that the 
turbulent wake would be instrumental in delaying electrical breakdown of 
the carrier gas. 



Preliminary studies revealed that a corona discharge device would 
provide the most suitable means of ion injection because of the relatively 
high (atmospheric) pressures involved in the air flow in the cylinder 
wake ' . The characteristics of a corona discharge which make it 
a favorable injector are given below. 

One type of corona can be created when a sufficient voltage is placed 
across the gap between a circular ring and a centrally mounted needle tip. 
At voltages below that required to cause an arc discharge across the gap 
there exists a field of sufficient strength to ionize the gas in this 
region. The charges of polarity opposite to that of the tip are then 
trapped in a charge sheath about the needle tip. The effect of this 
sheath is to reduce the outer electric field strength. This fact when 
simply stated means that although a current still exists due to the migra- 
tion of charges of the same polarity as the needle toward the ring, the 
migration occurs in a field of weakened strength and these ions can be 
more easily removed from the influence of this field. Thus, particles 
of a desired polarity can be forced away from the electric field of the 
corona unit (See Figure I) . 



10 



1 1 . iOI.KC UL/.R ION ?XP£RIMi:NTS 

Preliminary work with molecular ions involved a study of the current 
vol tape characteristics of a corona discharge unit which consisted of an 
aluminum ring mounted in a teflon block and a needle tip shaped from an 
iron nail (see Figure II). Current-voltage (I-V) curves from this setup 
are shown in Figure II. It was learned from this test that any sharp 
edges in the ring will act as discharge points, and therefore would cause 
the unit to deviate from its desired operational etiarac teristics . It was 
also found that the position of the needle tip within the ring has a 
noticeable effect on the I-V characteristics. The test was also employed 
as a familiarization period on the safety requirements of high voltage 
work . 

From this point the study was divided into two phases. Phase one 
involved the design and study of a small corona unit to be incorporated 
with the aforementioned cylinder. The cylinder was mounted in a test 
section through which air flowed at velocities in the vicinity of three 
hundred feet per second. In this phase then, the corona unit was em- 
ployed to ionize air and the air flow would provide a carrier which 
would force some of the ions out from the inside of the ring and down- 
stream to a collector. 

For a drawing of the model chosen for this phase refer to Figure III 
The ring was designed to be of a size which would insure that the whole 
unit would be within the wake of a one centimeter diameter cylinder. 
Braces on the ring. attached it to the downstream side of the cylinder and 
the needle tip wire was placed through a hole drilled streamwise through 
the cylinder. Electrical leads were brought in through the test section 



11 



walls in such a manner as to provide contact with ring and needle at 

the ends of the cylinder. 

If the needle were positively charged, positive air ions would 
migrate toward the ring, which would be at ground potential. As stated 
earlier, the migration would occur in a region of weak field strength due 
to a sheath of negative ions held in the vicinity of the needle tip. 
Thus for this polarity it should be possible in theory to remove some 
of these positive ions from this field by means of the air flow before 
they reach the ring. Similarly, electrons could be generated and removed 
by simply reversing the polarity of the device. 

Initial tests were then made to determine I-V characteristics of the 
unit in stagnant air. These are shown for either polarity in Figure IV. 
The next series of tests involved taking I-V data with the cylinder and 
corona unit mounted in the test section with various air flow rates. 
The electrical and metering setup is shown in Figure V. At this point 
it was determined that our I-V data was not a true indication of what 
was happening within the corona unit. The wiring was crude at best and 
at higher voltages numerous corona discharges were occurring at several 
spots where sharp edges were exposed, providing parallel current paths 
and thus giving misleading data. In an attempt to remedy this situa- 
tion, all leads were brought into two spherical brass junctions. Wher- 
ever any sharp edges were present an attempt was made to round them off. 
The I-V data then taken are shown in Figure VI. For a detailed report 
on the operation of the unit as a generator (with collector current data) 
and effectiveness of design innovations refer to reference 6. 



12 



I IX. COULQIPAL ION SOURCE 

A. INTRODUCTION 

The second phase of .this sibudy entails the major part of the re- 
search carried out in 'the preparation of this thesis. This phase is 
concerned with devising a means of injection of ions of a much greater 
size than ionized air intto (the .■air flow downstream from the cylinder. 

An important consideration in the design of an effective EGD genera- 
tor is the mobility of the charged particleB which are to be forced out 
to the collector. Mobility is a measure of the velocity of a charged 
particle under the influence of an electric field. A high mobility in- 
dicates that it would be difficult to force the ion6 away from their migra^ 

tion path from the needle tip to the ring and therefore a low mobility is 

4 
desirable. In a study carried out for the Aerospace Research Labora- 
tories published in 1964 s , it was repeated (that particle size is an im- 
portant factor governing the [mobility of the ions. As shown below, 
particles on the order iof !K0 metiers tfco 10 meters (one micron) in dia- 
meter provide a somewhat optimal, mobility for EGD considerations. 

A comparison between the usefullneBs of molecular ions and larger 
sized ions can be made using ttJhe Tndbilitty (calculations of reference 4. 
Consider the drift velocity uof an loon charged im a corona device under 
an electric field of 3..0 « M) volts//m '(breakdown potential (K) of air 

at STP). The respective ftrfljftt vva&ecities ( 1$~J) are given below. 

/*/ -4 2 

mw>llecul»r ion:: mebility ( JUL ) — 10 m /volt -sec 

-TVh =J*\~ 3.t) x 1'0 2 m'/secs 9.84 x 10 2 ft /sec 

*w -6 2 
micron si'Zed ion: JUL— 10 m /volt-sec 



<fc/- =yUX-=. 3.Dtnv/sec « 9.84 ft/sec 



13 



As can be seen, the molecular ions can easily attain drift speeds 
greater than that of the main flow and little interception can be ex- 
pected; whereas the micron sized ions drift relatively slowly and can be 

intercepted more readily by the flow. The situation improves further with 

-8 -7 
particle sizes between 10" and 10 meters. 

Having determined the desired size for the charged particles, and 
acting on the decision to retain the corona as a source of ion genera- 
tion, it was then necessary to choose a means of producing charged parti- 
cles which would be of the order of one micron in diameter. Three methods 
of producing micron-size particles were considered and are described be- 
low. The method employing saturated steam was chosen. 

Method one involved the use of powder which is manufactured to have 
an average particle size of one micron. The powder could then be intro- 
duced to the air flow upstream of the cylinder and channeled through the 
corona unit. This is perhaps the simplest method and guarantees the 
proper particle size, but it does have the disadvantages of both the neces* 
sity of a large supply of powder and the discomfort of exhaust powder from 
the open system, which could provide a hazard to metering operation as 
well. Furthermore, it is difficult to produce homogeneous charged parti- 
cle distributions. 

The second method considered was to replace the needle tip of the 
corona with a hypodermic needle. In this way a dielectric fluid could 
be forced out through the needle at a controlled rate. As droplets of 
dielectric form at the needle tip, the high electric field strength 

charges them and subsequently causes them to break up. Unfortunately, 

9 
there is no good way with this method to insure proper particle size. 

Finally, a method has been studied whereby saturated steam is con- 

2 
densed by means of expansion through a nozzle. The dry steam would 



14 



initially be forced through a corona unit as close as possible to the 
point of condensation. Then, in theory, the steam would saturate and 
condense about the ions, forming a colloidal suspension of charged water 
droplets in air. The steam is employed both as the dielectric and as 
the carrier gas to force the ions out of the nozxle. This method was 
chosen as the best suited for the work which haa been done here although, 
admittedly, it also presents the problem of how to grow the proper ion 
size during condensation. 

B. DESIGN OF COLLOIDAL ION GENERATOR 

A design was sought which would incorporate the cylinder with a 
steam supply, a nozzle and a corona device. The design chosen involved 
a hollow teflon cylinder fitted with a stainless steel nozzle which would 
face the downstream direction of the flow. In this configuration the 
nozzle could serve a three fold purpose. It would act as a flow metering 
device, yielding any desired flow rate based on a supply of saturated 
steam (i.e., 250°F. and 15 psig) . Additionally, it would serve as the 
ring of the corona unit and finally serve to condense the steam about the 
charged ions through expansion at the nozzle exit plane. A diagram of 
the nozzle and cylinder is shown, in Figure VII. Sample calculations and 
principles for nozzle design are shown in the Appendixes. Three nozzles 
were constructed to have steam flow rates of .01 lbm/sec, .005 lbm/sec 
and .001 lbm/sec, all with a choked steam flow at the exit plane. 

In order to have a generator capable of producing a constant supply 
of steam at the desired temperature and pressure it was decided that a 
sealed and reinforced aluminum container would be the most desirable. 
A pressure cooker served nicely for this purpose. The generated steam 



15 



would then be transported to the cylinder and nozzle by means of ^-inch 
O.D. stainless tubing, A ball valve was placed into this line to serve 
as an on-off valve. Any excess pressure would be bled off by means of a 
needle valve installed on a length of tubing which led away from the main 
line. The total length of tubing from the pressure cooker to the cylinder 
was wrapped in heating tape to prevent condensation of the steam while 
it was in the line (see Figures VIII-IX). A pressure gauge was installed 
in the top of the cooker in order to provide a guide for pressure regula- 
tion. 

Initial testing of this new unit was solely concerned with perfecting 
as much as possible the steam generation aspect of the system. A 4000 
watt hot plate was found, which easily maintained a 15 psig pressure head. 
After completely insulating the top and sides of the pressure cooker with 
a layer of Sauereisen number 31 cement of approximately %-inch in thick- 
ness and then another %-inch thick layer of asbestos tape, condensation 
problems were solved well enough to maintain a steady flow of dry steam 
through the nozzle and thereby prevent any arcing of the corona due to 
spurts of condensed water. A length of twenty-thousandths-inch platinum 
wire was brought in from the upstream side of the cylinder and centered 
within the nozzle to serve as the needle tip of the corona unit (see 
Figure VII). 

C. EXPERIMENTAL PROCEDURE 

As previously mentioned, before any reliable electric data could be 
obtained, it was necessary to insure that a steady and dry steam flow 
would be emitted from the nozzle. To achieve this end the steam line 
was purged with a jet of high pressure air prior to each test. When 
this was accomplished the ball valve which separated the nozzle from the 



16 



pressure cooker was closed to seal off any steam flow in that direction. 
The four-quart pressure cooker was then filled to approximately three 
quarters of its capacity with distilled water and the hot plate was turned 
on to the maximum heating position. 

It took about twenty minutes for the water to begin to boil. During 
this interim the heating tape, which was connected on line with a rheostat, 
was maintained at 400°F. to ensure that the inner tubing temperature would 
be greater than the temperature of the steam and thus aid in preventing 
any condensation in the line. 

Initial data taking involved a set of I-V curves on the nozzle and 
needle as a corona unit. During these tests there was no steam flow 
through the nozzle. A 9et of these data is shown in Figure X. These 
tests were usually carried out during the period when the distilled water 
in the pressure cooker was being raised to the boiling point. 

As soon as the water began to boil at a substantial rate all valves 
were closed until the internal pressure in the pressure cooker reached 
15 psig. When this point was reached the bleed off needle valve was 
opened just enough to maintain this pressure. This action was continued 
for several minutes in order to insure that the water had reached its 
boiling point at the higher pressure. The ball valve was then opened to 
allow the steam flow to go up the line and through the nozzle as the 
needle valve was readjusted in order to steady the internal pressure at 
a constant 15 psig. It was also necessary to increase the rheostat set- 
ting, since the steam flow had a considerable cooling effect on the tubing. 
A heating tape temperature of about 350°F. (measured by a thermocouple 
mounted between the ' tape and the tubing) was found to be sufficient to 
avoid condensation. 



17 



When a steady steam flow was obtained I-V data were again taken on 
the corona within the nozrle. These data are shown in Figure XI. Electri- 
cal connections and metering setups are shown in Figure XII. At this 
point all systems were ready for a test of the unit as a generator (see 
Figures XIII, XIV and XV). 

An iron nail, polished and sharpened, was insulated by plexiglass 
and mounted on a traversal mechanism. The tip of the nail served as the 
collector unit. It was wired to a microammeter and then to ground. Tests 
were then run measuring the corona voltage, corona current and collector 
current for both polarities on the corona. The collector position down- 
stream of the nozzle was a variable as was the corona voltage. These 
tests were run at first using only the steam flow to force the charged 
particles out to the collector. A later test was made to determine the 
effect of the air flow in conjunction with the colloidal ion injection. 
Data are presented on Figures XVI, XVII and XVIII. Results and the ef- 
fectiveness of system modifications are discussed in the Analysis of 
Results section of this paper. 



18 



IV. ANALYSIS OF RESULTS 

The first test of the complete system proved rather discouraging 
as there was no measurable current flow from the collector. Two modi- 
fications in the system were simultaneously devised. A digital multi- 
meter, capable of measuring down to one hundredth of a microampere, 
replaced the Simpson microammeter (see Figure XV). Secondly, it was 
deemed that with the configuration of the ion generator it would be 
advantageous to devise a means of aiding the process of getting the 
colloidal ions out into the region of the collector. As a solution of 
this problem the interior of the nozzle was coated with a thin layer 
of dielectric epoxy, with only a narrow band of steel at the throat of 
the nozzle left exposed, (see (Figure XIX). With the tip of the center 
needle recessed back into the nozzle, this would then align the direc- 
tion of the electric force field toward the nozzle throat. Finally, 
to prevent the collector from acting as a discharge point when placed 
too near to the nozzle, a -set of diodes was connected in series be- 
tween the collector and the multimeter. With these improvements a mea- 
surable, though admittedly small, collector current was drawn, (see 
Figures XVI, XVII, and XVIII). 

Three primary observations were made from the data thus obtained. 
First of these is that the ion density was primarily a function of 
streamwise distance from the nozzle exit plane. Positioning of the 
collector in various points off center of this axis had little effect 
on the current reading. Secondly, the collector current reading for the 

negative ion case was about double that for the positive ion case. 

2 
Though consistent with the results of other researchers, the reason 



19 



for this phenomenon remains to be uncovered. Thirdly, colloidal ion 

suspensions were shown to be considerably more effective in the genera- 

6 
tor operation than higher mobility air ions. 

The epoxy coating on the inside of the nozzle was helpful, but it 
was a crude solution at best. The ineffect of corona voltage increase 
on collector current may indicate a breakdown of the dielectric epoxy in 
this region. This hypothesis is based on the observation that at lower 
voltages on one test a significant self induced oscillation of corona 
current produced a proportional oscillation of collector current, (see 
Figure XVII). While on the same run at higher corona voltages and there- 
fore increased corona current, the collector current was essentially 
constant, (see Figure XVIII) . These higher voltage currents were far 
more stable than lower readings, but the point to be observed here is 
that the increased current did not proportionally increase the collector 
current as it had at lower voltage values. 

Results given on this study are all for the nozzle designed to yield 
a flow rate of 0.001 lbm/sec. Use of the larger nozzles was attempted, 
but the present steam generator was not capable of maintaining a pres- 
sure head at these nozzles. A heating method with a greater BTU capa- 
city and improved insulation would be required to overcome this hurdle. 

The addition of air flow through the test section detracted from 
rather than aided the performance of the system. Admittedly little work 
was done with this configuration in air flow due to the time spent in 
perfecting the steam generator and achieving a non-air flow collector 
current. It is very possible that the injector design simply works 
better by relying only on steam for a carrier gas and expansion for 
condensation. It may also be that there is an air flow rate which 
would be favorable to the process. This study has yet to be carried out. 



20 



The data presented here must be interpereted taking into account 
the fact that the existence of micron sized particles was not proven by 
this study. Further, if they did exist, no attempt has been made to 
determine at what point in the flow they were condensed. 

Finally, no leakage of current between the injector and the collector 
was anticipated. Therefore, no test of the generator unit was made with- 
out steam or air flow. The possibility of this leakage should, however, 
be considered in an analysis of this data, 



21 



V. CONCLUSIONS 

The injector designs which were sought at the start of this project 
were built and successfully operated. A discussion of efficiency by 
usual standards at this point would be rather premature due to the some- 
what limited state of development of the hardware. Positive, though not 
conclusive, results were obtained. It is hoped that these results prove 
helpful to future studies. 



22 



VI. RECOMMENDATIONS 

The design of the colloidal ion generator should be reviewed. It 
would seem advisable to design a system which would employ the air flow 
to carry the colloidal ions out of the corona region, rather than rely 
so heavily on the steam flow. The use of the nozzle should be maintained, 
but an improved system should include a way to feed flowing air into the 
nozzle entrance plane without prematurely condensing the steam. 

The sys em which has been used in this research could also be im- 
proved upon enough to bring about a marked increase in collector current. 
The first improvem it needed here is to get a smooth dielectric coating 
on the inside of ie nozzle, such as a thin teflon cone. This will 
'.essen the doubts raised concerning the breakdown of the rough epoxy 
coating used in this work. Secondly, an increased steam flow rate would 
be greatly helpful in getting the colloidal ions out to where they can 
be gathered by the collector. The nozzles for this purpose have already 
been built. All that remains to be done is to generate steam fast enough 
to maintain the desired pressure head. Finally, a study with air flow 
should be made to determine the compatibility of this present system 
with original design criterion. 

It is also- noted here that there has been no attempt made as yet to 
determine at what distance from the nozzle exit plane the micron sized 
ions are condensed, if they existed at all. Several studies have been 
made of the rates of condensation which may prove useful to an investi- 
gation of this aspect of the EGD generator. Two of these are listed in 
the bibliography for future reference ' 



23 



-MOBILE IONS 




FIGURE I GENERATOR SCHEMATIC 



I'A 




RING 



2.251* 




RING 



NEEDLE 



TUirn 



70- 



60-- 



50-- 



<^40 



LJ 

o 



30-- 



20-- 



10- 



C0R0NA CURRENT 
VS 
VOLTAGE 

BASIC DESIGN 



NEG NEEDLE 





POS. NEEDLE 



•)— I 1 



9 tO II 12 13 14 15 16 17 18 
VOLTAGE (DCKV) 



19 



FIGURE E PRELIMINARY CORONA DEVICE 



25 



_J 

LU 

o 

o 

z 

or 






<r 


0. 
UJ 


en 

UJ 

n 


o 


£_> 


<t 


o 


o 


_l 


m 


(I 


LL 


o 


o 


C\J 



z 
or 



°l 

C j 

cr 1 

<*' 

j 

o 

X 

UJ 

-I 

Q. 

Q 

toko 



Nil 



LU 

_J 
< 

o 
c/> 



UJ 
O 

> 
UJ 

Q 




< 

_j 

Z> 
O 
UJ 



UJ 

o 

Li- 



ar 

Q 

O 

CM 
O 



Q 

-12 



Z(> 




5.0 
VOLTAGE (DCKV) 



FIGURE JZ CORONA CURRENT VS VOLTAGE 
MOLECULAR ION UNIT 



27 




RING 



FIGURE I ELECTRICAL SETUP OF 
MOLECULAR ION INJECTOR 



28 



SO- 



SO- - 



70- 



60- 



50- 



< 



40-- 



UJ 

tr 

5 30+ 

o 



20-- 



10- - 



20 



FIGURE 21 



MOLECULAR ION UNIT 
IMPROVED WIRING 
NO AIR FLOW 



NE6. NEEDLE 



ARC 




4jO 5.0 6.0 

VOLTAGE (DCKV) 

CORONA CURRENT VS VOLTAGE 
MOLECULAR ION UNIT 



29 




cr 
o 

i- 
< 
cr 

LxJ 
UJ 

o 



< 



o 
o 



UJ 

3 
O 



O) 



30 




Id 

< 
-I 
0- 



o 

< 
UJ 

o 

CO 

<r 
o 

i- 
< 
cc 

UJ 

z 

UJ 

o 



< 

UJ 

I- 
co 



M 

Ul 

<r 



31 




(i) VIEW OF STEAM UNIT IN TEST CHANNEL 




)»•••' c ••«•••* 




(ii) VIEW OF PRESSURE COOKER 



FIGURE IX 



32 



NOZZLE 


AS 


CORONA RING 


NO 


STEAM FLOW 


POSITIVE 


NEEDLE 




2 3 4 

VOLTAGE (DCKV) 



FIGURE X CORONA CURRENT VS VOLTAGE 
MICRON INJECTOR UNIT 



33 



40t 



30" 



^ 
^ 



20«- 



UJ 
O 



10- 



MICRON INJECTOR UNIT 
STEAM FUOW 



NE6. NEEDLE 




2 3 4-. 

VOLTAGE (DCKV) 



FIGURE XI 



CORONA CURRENT VS VOLTAGE 
MICRON INJECTOR UNIT 



34 




S* H 



< 
UJ 



3 

a: 



o o 

UJ 

-I "9 

< ? 

o 

H- O 

UJ o 

UJ i 



UJ 

q: 



o 
u. 



35 




FIGURE m GENERATOR UNIT IN TEST SECTION 







FIGURE XGf POWER SOURCE AND VOLTMETER 











. , air rr* 



' 





\— ^-*^s*^*4 




i 




FIGURE 221 MULTIMETER AND DIODE BANK 



37 



0.40- - 



0.30- - 



< 



0.20-- 



0.10- - 



5.2 KIL0V0LTS DC 
VARIATION APPROXIMATELY 
THE SAME AT OTHER 
VOLTAGES 



•• 



+ 



0.25 0.50 

L (INCHES) 



0.75 



.00 



FIGURE X2T GENERATOR CURRENT VS COLLECTOR 
DISTANCE FROM NOZZLE EXIT 



38 



0.3O- 



5.0 KIUOVOLTS DC 
RECORDED DURING I ( 
OSCILLATION 



0.20- - 



< 

=1 



CD 



0.10- - 




+—I *— I- 



H I— « k- 

I 2 3 4 5 6 7 8 

I C (/^A) 



-I 1- 



+ 



+ 



10 II 12 



FIGUREXSir GENERATOR CURRENT VS CORONA 
CURRENT - MCRON INJECTOR UNIT 



39 



0.40- • 



L » 0.15 INCHES 




VARIATION AT OTHER 


V 


APPROXIMATELY THE 


SAME 



0.30- • 



0.20- - 



< 



0.10- - 



< 

o 

b 



< 


< 


< 


< 


^L 


5- 


^L 


^. 








IO 


CVJ 


ro 


m 


10 



+ 



+ 



+ 



5.0 5.2 5.4 5.6 

VOLTAGE (KVDC) 



FIGURE 



GENERATOR CURRENT VS CORONA 
VOLTAGE - MCRON INJECTOR UNIT 



40 



NOZZLE 




FIGURE XIX EPOXY COATING ON NOZZLE 



41 



APPENDIX A 



CALCULATION OF NECESSARY STEAM 



FLOW RATE 



Basic assumptions: 

-4 '* 

1. maximum desired current = 10 ampelpes 

2 

2. average charge per ion = 10 "electron charges 

3. micron sized ions 

. -4 -4 
1 amp ■ 1 coulomb/sec «« 10 amp = 10 coulomb/sec 

-19 
1 electron charge = 1.6 x 10 coulomb 

•\ 1.6 x 10 coulomb /ion 

-4 
It follows that, for a current of 10 amp: 

ions req'd 10 coulomb ion „______ 

sec sec . m"17 , u 

1.6 x 10 coulomb 

12 
■ 6.25 x 10 ions/sec 

-13 3 
Volume of 1 micron particle = 5.25 x 10 cm 

3 
Density of water = 2.204 lbm/cm (liquid) 

»\ Flow rate for 1007 o charging efficiency a 

6.25 x 10 12 ions/sec X 5.25 x 10" 13 cm 3 /ion X2.2 x 10~ 3 lbm/cm 

= 7.25 x 10~ 3 lbm/sec 
With this flow rate as a basis, steam mass flow rates were chosen which 
would bracket the theoretical value. 



42 



APPENDIX B 



CALCULATION OF NOZZLE EXIT AREAS 



Assumption: 

1. Superheated steam may be approximately treated as an ideal gas 
The analysis below represents a preliminary estimate of the exit areas. 



Ideal p,ns equation used ; 



-V," 



k+1 



P 
o 

r T~ 
o 



\AJ max , •'•' R \ k + 1 / 

wh e re : 

W = mass flow rate lbm/sec 

A* = nozzle exit area for choked flow 

k = c /c for steam = 1.33 
P v 

R = 85.6 ft-lbf/ R-lbm 

P = 30 psi 
o 

T = 710° R 
o 

Substituting into the above equation, the following exit areas were 

obtained for each selected flow rate: 

1. W = 0.01 lbm/sec, A = 0.02157 in 

2. W ■ 0.005 lbm/sec, A = 0.01079 in 2 

2 

3. W - 0.001 lbm/sec, A = 0.002157 in 

Nozzles were designed with the following guidelines: 

1. Add 107 o to exit area for real gas effects 

2. Nozzle half angle to be approx. 30°. 

3. Max nozzle length to be approx. 1 cm. 

4. Max entrance to exit area ratio ■ 20. 



43 



The nozzle dimensions are shown in Figure B-l. No attempt has been 
made as yet to determine what the actual flow rate for each nozzle is 



44 



I. W=O.OI Ibm/sec 



0.I74-. 



2. W =0.005 Ibm/sec 




0.123-1 



3. W= 0.001 Ibm/sec 



r 



0.550-. 



0.213 



L 




0.246 



T 



ALL DIMENSIONS 
IN INCHES 
SCALE: w 



FIGURE B-I NOZZLE DIMENSIONS 



45 



BIBLIOGRAPHY 



1. Smith, J. M. , Electrodynamic Power Generation-Experimental Studies, 

Space Sciences Laboratory Report R62SD27, March 1962. 

2. Marks, A., Experimental Investigation of a High Potential, High 

Voltage Electrogasdynamic Generator, Marks Polarized Corpora- 
tion Report Now-0582-c, December 1967. 

3. Marks, A., Barreto, E. , Chu, C. K. , Charged Aerosol Energy Converter, 

AIAA Summer Meeting Report 63-158, June 17-20, 1963. 

4. Kahn, B., A Continuation of the Basic Study of Slender Channel 

Electrogasdynamics, Aerospace Research Laboratory Report 
65-4, January 1965. 

5. Gourdine, M. C, Study of MHD and EHD Free Convection Energy 

Converters, Plasmadyne Corporation Report ASD-TDR 62-320, 
June 1962. 

6. Wallace, D. W. , Molecular Ion Electrogasdynamic Flow Channel , 

M. S. Thesis, Naval Postgraduate School, Monterey, California, 
1969. 

7. Cobine, J. D., Gaseous Conductors , p. 248-281. McGraw-Hill, 1941. 

8. Shapiro, A. H., The Dynamics and Thermodynamics of Compressible 

Fluid Flow , p. 85, Ronald Press, 1953. 

9. Pfeifer, R. J. and Hendricks, C. D., "Parameter Studies of Electro- 

gasdynamic Spraying", AIAA Journal , v. 16, p. 496, March 1968. 

10. Hirth, J. P. and Pound, G. M. , "Condensation and Evaporation, 

Nucleation and Growth Kinetics", Progress in Materials 
Science , v. 11, MacMillan, 1963. 

11. Hill, P. G. , "Condensation of Water Vapor During Supersonic 

Expansion in Nozzles", Journal of Fluid Mechanics, v. 25-3, 
p. 593-620, 1966. 



46 



INITIAL DISTRIBUTION LIST 



No. Copie s 



1. Defense Documentation Center 20 
Cameron Station 

Alexandria, Virginia 22314 

2. Library 2 
Naval Postgraduate School 

Monterey, Califcrria 93940 

3. Commander, Naval Air Systems Command 1 
Attn: Mr. Milton A. Knight (Code AIR-34001) 
Department of thr Navy 

Washington, D. C. 20360 

4. Professor 0. Biblarz 4 
Department of Aeronautics 

Naval Postgraduate School 
Monterey, California 93940 

5. Chairman, Department of Aeronautics 1 
Naval Postgraduate School 

Monterey, California 93940 

6. LTJG William T. Ober III, USN 1 
76 Salem Street 

Andover, Massachusetts 01810 



47 



Unclassified 



Security Classification 



DOCUMENT CONTROL DATA • R&D 

(Security cleeeillcatlon at title, body ol mbmtrmct and Indexing annotation null be entered when the o—rall rmporl la claeeltled) 



1 ORIGINATIN C ACTIVITY (Corporate author) 

Naval Postgraduate School 
Montarey, California 93940 



2a HC^OKT IICUKITV CLUIIFICtTIOI 



Unclassified 



26 GKOUF 



3 9EPORT TITLE 

Ion Injectors for Single- and Two-Phase Electrogasdynaaic Generators 



4 DESCRIPTIVE NOTES (Typa ol report and Indue I ve datee) 

Master's Thesis; June 1969 



5 AUTHORfS; (Laat name, it ret name. Initial) 

William Taylor Ober II 



6 REPORT DATE 

June 1969 



la TOTAL NO. OF PACES 



47 



7b. NO. OF REFS 
11 



tm CONTRACT OR GRANT NO. 



b PROJEC T NO. 



9« ORIGINATOR'S REPORT NUMBERfSj 



9b. OTHER REPORT uo(S) (A ny other number* that may be aeeigned 
thle report) 



10 AVAILABILITY/LIMITATION NOTICES 

Distribution of this document is unlimited 



11. SUPPLEMENTARY NOTES 



12 SPONSORING MILITARY ACTIVITY 

Naval Postgraduate School 
Monterey, California 93940 



13 ABSTRACT 

Systems suitable for the injection of ions into electrogasdynamic (EGD) 
generator devices were built and tested. The mechanism of injection was 
based on a corona discharge, whereby ions moving through an electric field 
can be intercepted by a gaseous flow. The intercepted ions are of one 
polarity, insuring selective ion injection. Two types of injector units 
were investigated. One was a molecular ion device which produced ions 
directly from the carrier gas, and the other created larger sised ions, 
resulting in an aeroaol flow. The latter consisted of passing saturated 
steam through a corona discharge and injecting it into an air stream. 
In order to aid the injection process, the wake of a cylinder in the air 
stream was utilized in both cases. Most of the work done here was devoted 
to the design and testing of the aerosol flow device. The degree of 
success was moderate. 



DD .!St 1473 



Unclassified 



49 



Security Classification 



Unclassified 



Security Classification 



KEY WO HDS 



Elect roga adynamic 

Ion 

Micron Sized Particles 

Generator 



DD F °o R v M473 back 



'I 01 -807-68?) 



50 



1 



Unclassified 



Secu; ! Classification 



APPENDIX D 



United States 
Naval Postgraduate School 




THESIS 



MOLECULAR-ION ELECTROGASDYNAMIC 
FLOW CHANNEL 

by 
David William Wallace 



June 1969 



Tiuii document kab bzzn appiovzd &oi pub&Lc kz- 
Izabz and baJUi; -Ua diAVu.buti.on am untunitzd. 



pages Dl - D56 



Molecular-ion Electrogasdynamic 
Flow Channel 

by 



David William Wallace 

Lieutenant (junior grade) , United States Navy 

B. S., United States Naval Academy, 1968 



Submitted in partial fulfillment of the 
requirements for the degree of 

MASTER OF SCIENCE 

IN 

AERONAUTICAL ENGINEERING 

from the 

NAVAL POSTGRADUATE SCHOOL 
June 1969 



ABSTRACT 

This investigation evaluates the operating characteristics of an 
EGD (electrog_asdynamic) generator system which utilizes air both as 
the carrier fluid and as the source of injected ions. The design and 
construction of a flow channel and a corona ion injector are discussed, 
the performance of the ion injector is examined, and the results of 
attempts to obtain work by EGD energy conversion are presented. The 
experimental results presented and discussed are in reasonable agree- 
ment with expectations . The high mobility of molecular ions inhibits 
the conversion process and only 0.5% of the ions were removed from 
the corona by the air flow. Suggestions for improvements on the 
present system and the design of an advanced system are made. 



TABLE OF CONTENTS 

I. INTRODUCTION 11 

II. PRINCIPLES OF OPERATION 13 

III. EXPERIMENTAL APPARATUS 17 

IV. EXPERIMENTAL PROCEDURE 21 

V. RESULTS AND DISCUSSION 23 

VI. RECOMMENDATIONS 2 7 
APPENDIX A - FLOW MEASUREMENTS 2 9 
APPENDIX B - FLOW DETERMINATION INSTRUMENTATION 33 
BIBLIOGRAPHY 52 
INITIAL DISTRIBUTION LIST 53 
FORM DD 1473 55 



LIST OF ILLUSTRATIONS 
Figure 

1 EGD Generator Schematic 35 

2 Corona Discharge Principle 3 6 

3 EGD System Wiring Circuit 37 

4 EGD System Schematic 3 8 

5 EGD System Flow Channel 3 9 

6 EGD System Flow Channel 40 

7 EGD System Flow Channel 41 

8 Compressor 42 

9 Cylinder Wake Turbulence Map 43 

10 Corona Unit Isometric 44 

11 Corona Unit 4 5 

12 Instrumentation 4 6 

13 Connector Sphere 4 7 

14 Corona Current vs Voltage 4 8 

15 Corona Current vs Voltage 4 9 

16 Corona Current vs Flow Rate 50 

17 Collector Current vs Voltage 51 



LIST OF SYMBOLS 



Symbol 



A' Percent of test channel cross-section within 
boundary layer 

b Barometric pressure (in. Hg.) 

D Orifice diameter (inches) 

h Pressure differential across the orifice (in. H^O) 
w 2 

I Corona current 
c 

I Collector current 

g 

K Dimensionless discharge coefficient 

KV Kilovolt (10 3 volts) 

M Mach number 

P, Absolute pressure at test section (psi) 
b 

P Absolute pressure at orifice (psi) 

AP Pressure differential of pitot-static tube (cm HO) 

q Rate of heat loss 

RMS Root-mean-square 

T Temperature (°R) 

T f Flow field temperature 

T Wire temperature 
w 

Uo, Freestream velocity (ft/sec) 

u Average velocity in the boundary layer (ft/sec) 

u Mass -mean velocity (ft/sec) 



V Corona voltage 
c 

Y Dimensionless expansion factor 



<x 



Dimensionless area multiplier 



3 
/O, Fluid density at test channel (lb/ft ) 

D 

3 
/O Fluid density at orifice (lb/ft ) 

— fi 
Uh Micro-ampere (10 amp) 



ACKNOWLEDGEMENT 
The author expresses his sincere appreciation to Professor Oscar 
Biblarz of the Naval Postgraduate School, Monterey, California, for his 
assistance and guidance, and to Mr. Pat Hickey, Naval Postgraduate 
School technician, for his many technical services. Special thanks are 
given to fellow student W. T. Ober, LT(j.g.), USN, for principally 
providing the corona ion injector. 



I. INTRODUCTION 

High voltage, low current power sources have not been exploited 
commercially due to their high cost and the generally bulky size of the 
necessary equipment. The most familiar apparatus of this type is the 
Van de Graaff generator which 'pumps' charges from ground to a col- 
lector device by means of a moving belt, thereby generating a high 
electric potential between the collector and ground. In 1959, W. E. 
Bennett pointed out that since a greater charge could be carried in a 
volume than on a surface (such as the belt of a Van de Graaff generator) 
and since the charge transfer rate could be much higher with a moving 
fluid flow, the fluidic Van de Graaff generator promised higher currents 
and greater efficiency. Such a device is variously termed an electro- 
gasdynamic (EGD) , electrohydrodynamic (EHD) , or electrofluiddynamic 
(EFD) generator, depending on the transport medium used, and is the 
subject of this investigation. 

As indicated by the title of this report, we have chosen to pursue 
the investigation of the electrogasdynamic type device, henceforth 

referred to as the EGD system. Although it has been shown by various 

3 
researchers that higher efficiencies and greater currents are obtained 

if micron-sized charged particles are injected into the system, we have 

confined ourselves in this project to molecular ions , using air both 

as the transport fluid and the source of charged particles. It should 



11 



be noted that this investigation is the preliminary stage of a more com- 
prehensive, long-term project being carried on by Professor Oscar 
Biblarz, Department of Aeronautics , Naval Postgraduate School. 

An experimental EGD generator system was fabricated and in- 
stalled in Building 2 30 at the Naval Postgraduate School, Monterey, 
California, utilizing a corona discharge to provide charged particles. 
A test program was carried out to determine the basic performance of 
the flow channel and ion injector, and to evaluate the molecular-ion 
EGD generator system as a whole. 



12 



II. PRINCIPLES OF OPERATION 

The electrogasdynamic generator is analogous to the Van de Graaff 
generator in principle: a moving medium transports charged particles 
against an electric field to a collector. Thus the kinetic energy of the 
transport medium is converted into electrical energy. In the case of 
the Van de Graaff generator, a moving belt is used to transport the 
charged particles. For the Van de Graaff generator, the amount of 

charge transported is limited by two physical criteria the surface 

area of the belt and the velocity of the belt. By using a moving fluid 
for the transport medium, the amount of charge transportable is greatly 
increased because: 1) the amount of charge carried is now more a 

function of volume rather than surface area; and 2) the fluid can travel 

2 
at a higher speed than a mechanical belt. 

An EGD generator consists of three basic components: 1) a 

charged-particle injector; 2) a dielectric conversion section; 3) a 

collector. Figure 1 is a schematic of an EGD generator system. 

Charged particle injection at near-atmospheric pressures is most 

4 
easily accomplished using a corona discharge . A high voltage is 

placed across a needle and ring, producing an intense electrical field 

in the region between the needle and the ring. Ionization of gas 

molecules in this region is caused by the electric field, with a sheath 

of oppositely charged particles being formed around the needle. This 

charge sheath intensifies the field adjacent to the needle, and diminishes 



13 



the field external to the sheath extending to the ring. Ions of like charge 
to the needle then drift toward the ring, as shown in Figure 2. The ion- 
ization process is exponential in nature, and produces electron 
avalanches which are a mechanism by which current flows across the 
corona gap. Each avalanch decreases the gap potential. Thus at low 
voltages, complete ionization across the gap is stopped by the asso- 
ciated potential decrease before breakdown can occur. At a sufficiently 
high voltage, however, the gap potential, even with the loss due to the 
electron avalanches, is sufficient to cause complete ionization and a 
direct current flow across the gap. Breakdown is undesireable because 
ions are no longer drifting alone towards the ring. 

In the case of a positive needle corona, gas molecules are 
stripped of an electron and drift towards the ring as positive charged 
particles. With a negative needle corona, the drifting particles may 
be free electrons or negative ions formed by electron attachment. Since 
both free electrons and negative ions have a higher mobility than 
positive ions, a negative needle corona produces higher current than 
a positive needle corona. As long as the breakdown potential of the 
gas is not exceeded, any current flowing through the corona unit will 
be due to ions reaching the ring and being neutralized. Ideally, if 
an external force could be applied which would remove all the charged 
particles drifting toward the ring before they reached it, thus preventing 
current flow across the corona, the net work done by the corona circuit 
power source would be zero. 



14 



The external force used to remove the charged particles from the 
corona field may be derived from the kinetic energy of a gas flow. For 
the ions drifting toward the ring to be removed by the air flow, the 
kinetic energy imparted to them by the flow must be greater than the 
electrostatic energy of the corona unit. In the case of molecular-sized 
particles this transfer of kinetic energy is accomplished by molecular 
collisions between the air flow particles and the charged ions, and 
thus is dependent on the mobility of the charged particles. (Mobility 
is defined as the average drift velocity per unit electric field, where 
the average drift velocity is related to the reduction of the ion mean 
free path by collisions with the gas molecules) . Mobility decreases 
as density increases, and therefore is also inversely related to 
pressure. Thus the removal of charged particles from the corona, 
which increases as mobility decreases, increases with the increasing 
pressure . 

Once the gas molecules, now charged, are removed from the 
corona, their electrical energy is retrieved by means of a collector. 
This may be a wire mesh, a hollow cylinder, or a conducting rod. The 

rod seems to be the most efficient method of collection, although it is 

6 
not understood exactly how the collection process occurs. 

The retrieval of charges at the collector causes a buildup of a 

potential between the collector and ground, and the resulting current 

flow through the collector circuit is the work done by the air flow, 

or the output of the EGD system. 



15 



Figure 3 shows the wiring circuit of an EGD system. Ideally, 
there will be no current flow through the corona circuit (I = 0) since 
the ions will be swept downstream to the collector, producing a collector 
current (I ) . 

With the corona -conversion process acting as a generator, a 
typical load curve might be obtained by placing the generator output 
across a suitable variable resistance. For a given resistance, a voltage 
versus current curve could be obtained which would include the maximum 
voltage-no current case (open circuit) and the no voltage-maximum 
current case (short circuit) as its limits. Since the conversion process 
at the corona was of primary interest, e.g. , the percentage of corona 
ions being forced to the collector, the short circuit setup which in- 
dicated largest current through the collector was first used. After a 
significant collector current had been detected, the generator perfor- 
mance would have been further evaluated by determining the entire 
load curve, had more time been available. 



16 



III. EXPERIMENTAL APPARATUS 

The EGD system fabricated was designed for approximate flow 
speeds of Mach 0.3 and for minimum flow turbulence at the test section 
where the corona unit was placed. Figure 4 shows the system schematic, 
while Figure 5 shows the flow channel in detail. A photograph and 
engineering drawing of the flow channel are provided in Figures 6 and 

7, respectively. Air was supplied by a Carrier three stage centrifugal 

3 
compressor (Figure 8) with a 4000 ft /min. maximum inlet capacity and 

a maximum pressure ratio of two. The exit air temperature varied from 
65°F to 240°F. The flow could be regulated, as shown in Figure 4, 
by two valves and exhausted to the atmosphere. The flow orifice shown 
in Figure 4 was used only to determine the accuracy of the pitot-static 
tube located in the flow channel, which in conjunction with the man- 
ometer bank was used to determine the flow rate setting (See Appendix 
A) . The cooling bank was used to keep the temperature of the air flow 
close to ambient and to hasten the temperature stabilization. An iron- 
constantan thermocouple was used to determine the flow temperature. 

The test channel was made of three-eighths inch Plexiglass which 

was chosen because of its high dielectric properties (volume resis- 

12 
tivity=10 ohms-cm). In addition, Plexiglass permitted the investi- 
gator to observe the interior of the test channel. A test channel plenum 
to test section area ratio of 10 to 1, and a series of consecutively finer 
honeycomb and wire mesh were used to reduce the freestream flow 



17 



turbulence to a level comparable to contemporary flow channels — 0.14% 
RMS (i.e. , the time-averaged magnitude of velocity fluctuations was 
0.14% of the freestream velocity) according to a Ballantine RMS volt- 
meter. Using a Security Associates single-channel hot-wire probe, the 
channel boundary layers were determined so that the corona unit might 
be placed outside of the boundary layers. The boundary layers were 
found to be not fully developed, and at most only 0.15 inch thick which 
presented no interference with the test region. 

Because the efficiency of a corona discharge depends in part on 
pressure, being more efficient at lower pressures, the corona unit 
was mounted on the downstream side of a Plexiglass cylinder placed 
horizontally in the flow channel. Thus, due to the pressure drop 
downstream of the cylinder, the pressure at the needle where the corona 
action is strongest was relatively low compared to the pressure at the 
ring, where the conversion process occurs. It was observed that the 
corona performance was also affected by turbulence in that turbulence 
tended to inhibit breakdown. Although this aspect of the corona unit 
was not investigated, turbulence levels in the wake of the cylinder 
were determined, and are given in Figure 9. The ring and needle leads 
were buried in the Plexiglass to minimize the possibility of point dis- 
charges and subsequent current leaks. Figure 10 is an isometric of 
the corona unit, while Figure 11 provides an engineering drawing of 
the corona device. The cylinder was three-eighths inch in diameter. 
The Reynolds number based on freestream velocity and cylinder diameter 

18 



4 
for M=0.3 was 6.63x10 , which is subcritical, indicating separation 

close to 90° from the horizontal centerline. The needle and ring were 

made of pure platinum wire with wire sizes of 0.010 and 0.020 inches 

respectively. Three different corona configurations were used. Their 

pertinent dimensions are summarized as follows: 

Distance From 



Ring 




Needle 
Length 


Cylinder to 
Ring (smallest) 


Ring 
Diameter 


1 

2 
3 


0. 
0. 
0, 


187 inch 
,062 inch 
. 187 inch 


0.187 inch 
0.187 inch 
0.187 inch 


0.3 94 inch 
0.315 inch 
0.750 inch 



The collector unit consisted of an iron rod (1/8" Dia.), sharpened 
to a needle point, which was mounted on an insulated traversal unit 
allowing vertical and horizontal movement for any axial position. 

High voltage power was supplied to the corona unit by a 
Sorensen High Voltage D.C. Power Supply. This power supply produces 
up to 30KV and 2 0MA of current with a ripple of 2%, and has trip controls 
for both voltage and current. It also has the feature of reversible 
polarity. 

The high voltage across the corona was measured with a Sensitive 
Research electrostatic voltmeter. This instrument has an internal 
impedance of 5x10 ohms and reads RMS voltages up to 40KV on four 

scales. 

Corona currents were measured with a Simpson 0-100 UK ammeter. 



19 



Collector currents were read using a Calico Digital Multimeter 
which had an amperage range of 0.01-1000 U A through the use of 
various scales. 

Figure 12 is a photograph of the experimental setup. 



20 



IV. EXPERIMENTAL PROCEDURE 

This investigation was divided into two parts: 1) design and con- 
struction of apparatus and determination of its airflow properties, and 
2) testing of the EGD system. 

The first part was preliminary to the second and consisted of 
determining and reducing, if necessary, the freestream turbulence; 
determining the test channel wall boundary layers; and exploring the 
corona -cylinder wake (See Appendix B) . Turbulence levels were 
determined using the RMS voltmeter in conjunction with the hot-wire 
anemometer, assuming the boundary layer to start where the velocity 
dropped to 0.99 of the freestream velocity. After the freestream turbu- 
lence was lowered to an acceptable level by means of honeycomb and 
wire mesh screen, and the boundary layers were determined, the corona 
cylinder was placed in the test channel. Turbulence levels and wake 
boundaries were determined in order to insure that the corona ring was 
placed within the wake region where high turbulence levels would 
inhibit breakdown. This completed preliminary work and actual testing 
of the corona unit was begun. 

The corona unit was first tested for continuity of all electrical 
connections. An effort was then made to minimize point sources which 
might enable current leakage. This problem was primarily attacked by 
devising the brass connector spheres shown in Figure 13. Then data 
for corona curves was taken without air flow, followed by data for 



21 



corona curves with air flow. Curves were obtained by holding the flow 
velocity constant and varying corona voltage, and by keeping corona 
voltage constant and varying flow velocity. Finally, the collector unit 
was positioned downstream of the corona unit. If a collector current 
was detected, the unit was moved about to determine the effect of 
position on collector current. Prior to each test run, the corona unit 
was cleansed with freon. This procedure was carried out for each of 
the three corona units . 



22 



V. RESULTS AND CONCLUSIONS 

Figures 14 thru 17 show the data obtained by the procedure out- 
lined previously. The general pattern is the same for each corona con- 
figuration (see page 19, this report) , and thus only curves for a single 
configuration are given. Data reproductability was generally good. The 
main cause of data error was improper circuit connection; specifically, 
if the contact screws were not seated properly, internal corona dis- 
charges would occur, causing noticeable errors. Small data variances 
from run to run were caused by the relative insensitivity of the ammeter 
in the corona circuit. The smallest scale division of this ammeter was 
2 U A and the meter itself had an accuracy no better than +2 WA. Other 
data discrepancies may have been due to humidity variations, mis- 
alignment of the corona needle, or power supply irregularities. 

Figure 14 is a typical comparison of corona current versus 
voltage with no airflow for positive and negative needle coronas, 
using corona unit 1. As shown, a negative needle corona produces 
much higher corona currents for a given voltage than a positive needle 
corona , and the breakdown of a negative needle corona greatly exceeds 
that of the positive needle. However, in subsequent data runs the 
positive needle corona was used since this polarity produces ions 
drifting toward the ring of a lower mobility than those produced by a 
negative needle corona . Thus greater collector currents would be 
expected with a positive needle corona. 



23 



The next chart, Figure 15, shows the results of adding airflow 
to the positive needle corona. At low test section Mach numbers 
(M <0.3), the corona current at a given voltage is considerably less 
and the breakdown potential is slightly higher than that of the no-flow 
corona. At higher Mach numbers (M> 0.3), the corona currents at a 
given voltage are only slightly less than that of a no-flow corona but 
the breakdown potentials have dropped considerably. It is believed 
that this behavior of the corona current first decreasing, then in- 
creasing as the velocity is increased at a given voltage is due to 

interaction between the velocity and pressure effects on corona 
efficiency. At low velocities and pressures velocity has a greater 
effect than pressure, thus causing decreasing current with increasing 
velocity. After a certain increase in velocity and accompanying 
decrease in pressure, however, the pressure effect becomes dominant 
over the velocity effect, causing an increase in corona current with 
further velocity increase. 

The current -voltage -velocity relationship was investigated with 

a different approach by holding the corona voltage constant and examining 

how the corona current reacted to changes in air flow rate. The results 

are shown in Figure 16. It can be seen that addition of flow rate causes 

an initial drop in corona current at any voltage. The corona current then 

increases as velocity increases, but tends to remain below the no-flow 

level. Both the initial drop and gradual increase are more pronounced 

at high voltages. Data points cross-plotted from Figure 16 to Figure 15 

lie on the existing data curves. 

24 



Efforts to detect collector current using corona unit 1 at any air 
flow rate were generally unproductive, and led to the designing of 
corona unit 2. This unit had the feature of a shorter needle but because 
of a smaller-radius ring the distance between the needle tip and the 
ring was the same as that of corona unit 1 . Thus the electric field 
between the needle and the ring would tend to help drive the drifting 
ions downstream. A main problem encountered was the action of the 
collector needle as a corona needle when placed close enough to the 
ring to detect generated current. Any collector current which may have 
been present was being overshadowed by the backward flowing current 
of the unwanted corona. This problem was eliminated by placing a set 
of three International Rectifier Corporation diodes in the collector 
circuit which prevented any reverse current. Since no collector 
current was yet detectable, another corona unit, unit 3, was built. 
This unit had a larger diameter ring which reduced the electric field on 
the charged ions, enhancing their chances of receiving enough kinetic 
energy to be blown downstream to the collector. Figure 17 shows the 
results of this final effort. Since no current was detected with corona 
units 1 and 2 in the same range of voltages using the diodes to prevent 
a reverse corona, the currents obtained were reasonably assured to be 
collector currents rather than reverse current leakage through the diodes. 
The bars representing data points are due to the fact that these currents 
were read from a digital ammeter on the last digit of the readout. Thus 
it was deemed inappropriate to average the values which fluctuated over 



25 



the range shown. Collector current increased with increasing voltage, 
being limited by the breakdown potential of the corona . The optimal 
collector position was on the corona ring centerline, one inch downstream 
of the ring plane. Any horizontal or vertical movement from this position 
caused a decrease in collector current. Changes in air flow rate seemed 
to make little difference in the amount of collector current, although as 
can be seen by Figure 17, the amount of current detectable was so 
minute that changes may have been unreadable due to limitations of the 
ammeter. The trend of the curve is reasonable in that increased voltage 
causes increased corona current. As corona current rises, more ions 
are available for conversion. 



26 



VI. RECOMMENDATIONS 

As stated earlier, for an EGD system to produce power, the force 
imparted by the gas flow on the ions of the corona must exceed the 
attractive force of the corona field. Since variation of the corona 
voltage is relatively restricted by breakdown, the most apparent 
improvement would be to increase the rate of energy transfer between 
the flow particles and the injected ions. On this basis, the following 
recommendations are made: 

a. Investigate further the pressure wake of the cylinder to 
re-evaluate its effect on the EGD process. The benefits to corona 
efficiency caused by the decreased pressure of the cylinder wake may 
be less than the detrimental effects on the conversion process, which 
is more efficient at higher pressures. 

b. Conduct further modified experiments by injecting micron- 
sized particles (such as dust) into the air flow upstream of the corona 
unit. Such particle ions have a much lower mobility than air ions and 
thus should lead to higher conversion efficiencies . 

c. Develop a new injection system which would inject micron- 
sized charged particles into the air flow. (Such a system might use 

steam as the injection medium. LT(jg) W. T. Ober, USN, has con- 

7 
ducted an investigation of a system of this type. ) These lower 

mobility ions would again increase the conversion efficiency. 



27 



d. Since ion mobility is reduced by increased pressure, the 
conversion process would be enhanced if the operating pressure 
(atmospheric for this investigation) could be increased. 

e. Continue efforts to improve data reproductibility by elim- 
inating sources of current leakage and other adverse system variables 



28 



APPENDIX A 
Flow Measurements 

The experimental facility air flow delivery tube had a square- 

g 

edged orifice, built to ASME standards, for determination of flow rates. 
However, the equation for mass flow as determined by this device is 
awkward and requires an iterative procedure to obtain a value. Since 
frequent flow rate changes were necessary for this investigation and 
extremely accurate flow rate determination was not a requirement, it 
was deemed sufficient to measure flow rate using a pitot-static tube 
located in the test section. However, the following calculations were 
made to ascertain how closely the pitot-static and orifice measurements 
correlated. 

The development of the orifice flow rate equation is , according 
to reference 8: 



2. 



W = 359.1 (D 2 ) Y 1 c<K /°i h w Gb/hr) (A-D 



D = orifice diameter, inches 

3 

/O = fluid density at orifice, lb/ft 



h = pressure differential across orifice, 

w 

inches H_0 



Y = dimensionless expansion factor 



K = dimensionless discharge coefficient 
o< = dimensionless area multiplier 



29 



K= 0.6876 (Ref. 8, p. 221) (A-2) 

A> may be determined as follows: 

/^ = MP 1 /(10.73)T 1 ^ 1 (A-3) 

T = temperature above orifice, °R 

P = absolute pressure above orifice, 

lb/ft 2 

M = 28 . 96 for air, 77" — , 

lb-mole 

U -1.0 (Ref. 8, Fig. 138) 

Y is obtained from the following formula: 

Y, = 1-(.41 + .35/5 4 )h /KP, (A-4) 

1 w 1 

/3 = orifice to pipe diameter ratio 

K = 1.4 for air 

Y , £ , andc<are all very close to unity. For D = 2.15 inches , the 
orifice flow rate equation reduces to: 



W= 1131 //> 1 h w (lb/hr) (A-5) 

or by the continuity equation: 

u = W/^> b A (A- 6) 

A = test channel cross-sectional area 
u = mass-mean velocity 
/° = fluid density in test channel 



30 



The equation for flow determination by the pitot tube is derived 
from Bernoulli's equation: 

p , + /° a U . 2/2 9 = Pk + /°k U k 2/2 9 ^" 7) 

a ' a a b'bb 

but U = , /3 = /O giving after rearrangement: 
a f a / b 



U^ = ^2gAP//O b (ft/sec) (A-8) 

g = 32.2 ft/sec 2 
AP = pitot tube pressure differential, cm HO. 

However, the orifice equation gives mass-mean flow rate while 
the pitot tube indicates freestream velocity. Therefore, the pitot 
tube value is corrected according to Reference 9, page 536. For a 
1/7-power velocity distribution, in the boundary layer, 

(u/U^j ) average = .875. 
Therefore, the mass -mean velocity can be computed by the following 
formula: 

u = [A' x u] + [(1-A 1 ) x Uoo] (A-9) 

u = average velocity in the boundary layer 
Uoo = freestream velocity 
A' = percent of cross-sectional area within 
the boundary layer. 



31 



Assuming a boundary layer thickness of 0.1 inch, A-9 reduces to: 

u = .2 8u + .72UOO (A- 10) 

or 

u = .965UQO (A-ll) 

The following data is reduced for comparison: 

Orifice Pitot Tube 

t = 69°F t 2 - 68°F 

h = 26.1cmH AP = 46.3cmH_0 

P = 86.6 cm HO P = 70.0 cm HO 

X £t £t La 

b= 30.06 in Hg b = 30.06 in Hg 

/o = .0819 [by (A-3)] /O = .0806 [by (A-3)] 

U^ =2 76 ft/sec [by (A-8)] 
u = 280 ft/sec [by (A-6)] u = 266 ft/sec [by (A-ll)] 

Thus flow rate determination using the pitot-static tube yields 
values approximately 5% in error on the low side. Assuming constant 
density for air, the following formula, which includes a factor to 
correct for this error, was used to compute flow velocities: 



u = 40.95 JAP (ft/sec) 
As before, &P is the pitot-static tube pressure differential in 
centimeters of water. 



32 



APPENDIX B 
Flow Determination Instrumentation 

Before choosing the means of determining the flow properties of 
the test channel, four different devices were considered: the hot-wire 
anemometer, thermistor, conductivity probe, and pitot-static tube. 
The thermistor and conductivity probes are devices which measure 
temperature and temperature fluctuations. The thermal sensitivity of 
the thermistor is about twice that of the conductivity probe (-4%/°C 
versus -2%/°C) but has a slower frequency response. However, both 
the thermistor and the conductivity probe are insensitive to velocity 
fluctuations. Inasmuch as the flow properties which were under 
investigation were velocity and velocity fluctuations (turbulence) , 
the thermistor and conductivity probe were immediately eliminated as 
useful. 

Thus it was decided to use the hot-wire anemometer which gives 
both the relative velocity (generally the output indicator is adjusted 
to maximum scale for a specific local velocity) and, in conjunction 
with a RMS voltmeter, velocity fluctuations. This device utilizes 
King's Law which gives the rate of heat loss from an electrically heated 
fine wire exposed to a flow: 



33 



q = (A + B fu) (T -T J where 
1 w f 

q = rate of heat loss 

T = wire temperature 
w 

T, = flow field temperature 

U = velocity normal to the wire 
A and B are constants dependent on geometrical factors and fluid 
property values. A hot-wire system may be operated in one of two 
modes: constant current or constant temperature. The constant temper- 
ature mode has a frequency response several orders higher than the 
constant current mode and thus has better response characteristics. 
In this mode a feedback amplifier is employed which senses the wire 
resistance and adjusts the wire heating current to maintain a constant 
wire temperature. Changes in flow velocity over the wire cause heating 
loss rate changes which are sensed by the amplifier as resistance 
changes. The hot-wire anemometer system used was a Security 
Associates Constant Temperature Hot-Wire Anemometer which utilizes 
a built-in analogue computer circuit to linearize the output. 

In addition, a pitot-static tube was used to measure the flow 
mean velocity. 



34 



CHARGED- PARTICLE 
INJECTOR SECTION 



I 



X 



COLLECTOR SECTION 







DIELECTRIC 
CONVERSION 
SECTION 



O ► UNCHARGED MOLECULES 



-► CHARGED MOLECULES (IONS) 



FIGURE 



EGD GENERATOR SCHEMATIC 



35 



CORONA RING 



CORONA NEEDLE 





CHARGE 
SHEATH 



HIGH VOLTAGE DC 
POWER SUPPLY 



:} 



CHARGED PARTICLES 



FIGURE 2. CORONA DISCHARGE PRINCIPLE 



36 



GAS FLOW 



CORONA 
VOLTMETER 




RING 




COLLECTOR 



COLLECTOR 
AMMETER 



CORONA 
AMMETER 



NEEDLE 
AMMETER 



HIGH VOLTAGE 
POWER SUPPLY 

(DC) 




LOAD 



- CHARGED PARTICLE 



FIGURE 3. EGD SYSTEM WIRING CIRCUIT 



37 




t 



< 



IS 

Q. 




UJ 

Q TT 



o 




I 



♦ 



_J 

UJ 



< 

X 

o 



o 

UJ 

X 

o 

CO 



UJ 

h- 
co 

>- 

CO 

Q 

e> 

UJ 



UJ 

cr 



CO 
UJ 

J- 



A 



7 



Q 
CD 
Ul 



38 




CO 

cr z 
oo 

CTcO 
t LU 



1 


<fr 


CVJ 


m 



LU 



< 

X 

o 

o 

-I 



2 

LU 

co 

>- 
co 

o 

CD 
LU 



LU 
CD 



CO 
UJ 



39 




FIGURE 6 EGD SYSTEM FLOW CHANNEL 



40 



in 
in 



o 

UJ 
CO 



UJ 
Q. 



O 
O 

b 



lO 






r^ 






ro 






i 






< 






IT 


CO 




UJ 


< 




h- 


-I 




< 


o 


UJ 


2 




X 


-1 


_i 


UJ 


< 


_i 


_J 


o 


< 


Q- 


<o 



1 



in 



O 
UJ 
CO 



co 

UJ 







i 

c 
c 

1 


I 



f 




UJ 

z 
< 

Q 



UJ 

»- 
CO 

>- 

CO 



o 

UJ 



UJ 

or 

e> 
u. 



i — r 



i 



c\i 



41 




FIGURE 8. COMPRESSOR 



42 



o <o o 
<fr ro «o 

(vi 01 - 



— '& 



s 2 88; 

0) - : ^ — ▼ 



3 5 2 



^ 00 

o f- & 



CM 2 £■ 

* - * 



O — o 
cm o> <VJ 














UJ 






CO 






V* 






K 






ll 






3> 






^- 






ro 







O " 8 




Id 


UJ 15 




CO 


CO . 




u. 


FT/ 
ING, 




I s - 


ro * 




*- 


* O 




ro 


ro \ 


r- 


11 


« £ 


z 


8 


8 




3 


=> o: 







LlI 


a. 


* 


- 




— 


CVJ z 


2 


h- 


t I] 


< 


z 


z > 


a 


3 


=> 




^p 


*"■» r— 1 


X 


0^ 
2 


J 2 




_^ 1 1 



«.=. 8? 



J^oo 

O CO IO 
CO CM K) 

IO .*..'». 

X 



28 § 

IO *• 0> 
X 

go'cT 
00 r^- cm 
00 CO o> 

X 



£ * £ 

en — r^ 



CO 

o 

CO 

o 

Q. 

< 

CO 

o 

or 

5 



mm s 



O CM 2 



— en — 



^ evj O 

* CO CM 



OK) t 
CM cCJ 



li- < 

O 

2 £ 



UJ 

o 



- ^ Ul 



o 
o 



%l= 



in to O 

CM N N 

co ^- to 



CVJ 



in 



o i° ° 

CM IS — 




3NHcJ31N30 IVlNOZIdOH 3A08V 



«q|p) 30NV1SIQ 



3 
03 

a: 



0> 
UJ 

U. 



43 




CO 

QC 
O 
CO 



< 



o 
o 

Q 

z 
< 



o 
a: 
o 
o 



UJ 



44 




Q. 

>- 








CO 




LlI 


z 




X 


Z> 




o 


z 




z 




£ 


z 


Q. 


£ 




O 
CM 
O 




CO 

o 


* 




CO 




z 




LU 




2 






o 



X 

-J 
< 
o 

CO 



o 
tr. 
O 
o 



LU 

tr 



45 




FIGURE 12. INSTRUMENTATION 



46 




FIGURI . CONNECTOR SPHERE 



47 



90-- 



80-- 



70-- 



< 



60-- 



tr 

§ 50- 

o 



o 
a: 
o 
o 



40-- 



30-- 



20-- 



10-- 



N0 AIRFLOW 
CORONA UNIT 



j 
i 



NEGATIVE NEEDLE 
• POSITIVE NEEDLE 
X BREAKDOWN 




50 6.0 7.0 8.0 
CORONA VOLTAGE (KV) 



FIGURE 14. CORONA CURRENT VS VOLTAGE 



48 



40-- 



30- ■ 



< 



UJ 

£ 20 

K 

o 

< + 

z 
o 

QC 
O 
O 10 




30 



CORONA UNIT I 
POSITIVE NEEDLE 
VARIOUS AIR FLOWS 




i 



• 
a 
x 

o 



U =0 FT/SEC 
U = 153 FT/SEC 
U * 239 FT/SEC 
U *334 FT/SEC 
= 363 FT/SEC 



4.0 5.0 6.0 

CORONA VOLTAGE (KV) 



70 



8.0 



FIGURE 15. CORONA CURRENT VS VOLTAGE 



49 



40-- 



< 



30- 



CORONA UNIT 1 




POSITIVE NEEDLE 




VARIOUS VOLTAGES 




• 


5.0 KV 


D 


5.5 KV 


X 


6.0 KV 


A 


6.5 KV 


O 


7.0 KV 



oc 
o 



o 
or 
o 
o 



20-- 



10-- 






Mr 



h h 

210 



H 1 1 h 



-i H 



H 1 1 h 



230 250 270 290 310 

FLOW VELOCITY (FT/SEC) 



330 



FIGURE 16. CORONA CURRENT VS FLOW RATE 



50 



10- 



C0R0NA UNIT 3 
POSITIVE NEEDLE 




70 8.0 

CORONA VOLTAGE (KV) 



FIGURE 17. CORONA AND COLLECTOR 
CURRENT VS VOLTAGE 



51 



BIBLIOGRAPHY 

1. Bennett, W. E. "The Generation of Direct Current at High 
Potentials." Research Applied in Industry , Vol. 12, No. 12, 
December 1959, England. 

2. Smith, J. M. Electrohydrodynamic Power Generation ' 

Experimental Studies. General Electric Space Sciences 
Laboratory Report, March 1962, 

3. Decaire, Capt. John A. and Wifall, Capt. James R. "Charge 
Generation by Corona Discharge in Electrofluiddynamic 
Conversion Processes . " Advances in Energy Conversion 
Engineering . ASME, 1967. 

4. Charged Particle Power Generation and Propulsion. Maremont 
Corporation, Final Report on Contract NOw 64-0594 -f, Bureau 
of Naval Weapons , Washington, D. C., July, 1966. 

5. Cobine, J. Gaseous Conductors . McGraw Hill Book Company , 

6. Marks, A. and Barreto, E. "Charged Aerosol Energy Converter. " 
AIAA Tournal , Vol. 2, No. 45, January 1964. 

7. Ober, LT(jg) William. Ion Injector for Single and Two-phase 
Electrogasdynamic Generators . Master Thesis, Naval Post- 
graduate School, Monterey, California. June, 1969. 

8. - Stearnes , R. F . , et al . Flow Measurement with Orifice Meters . 

Van Nostrand Co. , Inc., New York, 1960, 

9. Schlichting, Hermann. Boundary Layer Theory . McGraw-Hill 
Book Co., Inc., New York, 1960. 

10. Gibson, C. H., Chen, C. Co, and Lin, S. C. "Measurements 
of Turbulent Velocity and Temperature Fluctuations in the Wake 
of a Sphere." AIAA Journal , Vol, 6, No. 4, 1968. 



52 



INITIAL DISTRIBUTION LIST 



No. Copies 



1. Defense Documentation Center 2 
Cameron Station 

Alexandria, Virginia 22 314 

2 . Library 2 
Naval Postgraduate School 

Monterey, California 93 940 

3. Commander 1 
Naval Air Systems Command 

Department of the Navy 

Attention: Mr. Milton Knight, Code AIR-340C 

Washington, D. C. 20360 

4. Professor O. Biblarz 4 
Department of Aeronautics 

Naval Postgraduate School 
Monterey, California 93 940 

5. LT(jg) D. W. Wallace, USN 1 
2207 Jameson St. , S.E. 

Washington, D. C. 20031 

6. Chairman, Department of Aeronautics 1 
Naval Postgraduate School 

Monterey, California 9394 



53 



Security Classification 



DOCUMENT CONTROL DATA -R&D 



,Security classt lication ol llllo. body ol abstract and indexing annotation must be entered when the o vara /J report It i hissllled) 



ORIGINATING activity (Corporate author) 

Naval Postgraduate School 
Monterey, California 9394 



2a. REPORT SECURITY CLASSIFICATION 



Unclassified 



2b. GROUP 



3 REPORT TITLE 



Molecular-ion Electrogasdynamic Flow Channel 



4 DESCRIPTIVE N O T E S (Type ol report and. inclusi ve dates) 

Master's Thesis; June 1969 



5 authoRiSI (First name, middle initial, last name) 



David William Wallace 



6 REPORT DATE 



June 1969 



7a. TOTAL NO. OF PAGES 



52 



76. NO. OF REFS 



10 



8a. CONTRACT OR GRANT NO. 



b. PROJEC T NO. 



9a. ORIGINATOR'S REPORT NUMBER(S) 



9b. OTHER REPORT NOISI (Any other numbers that may be assigned 
this report) 



10 DISTRIBUTION STATEMENT 



Distribution of this document is unlimited 



I. SUPPLEMENTARY NOTES 



12 SPONSORING MILI TAR Y ACTIVITY 



Naval Postgraduate School 
Monterey, California 93940 



13. ABSTRACT 



This investigation evaluates the operating characteristics of an EGD 
(electrog_asdynamic) generator system which utilizes air both as the carrier fluid 
and as the source of injected ions. The design and construction of a flow channel 
and a corona ion injector are discussed, the performance of the ion injector is 
examined, and the results of attempts to obtain work by EGD energy conversion 
are presented. The experimental results presented and discussed are in reasonable 
agreement with expectations. The high mobility of molecular ions inhibits the 
conversion process and only 0.5% of the ions were removed from the corona by 
the air flow. Suggestions for improvements on the present system and the design 
of an advanced system are made. 



DD 



F0RM 1473 

1 NOV 65 I "T I *J 

S/N 0101 -807-681 1 



(PAGE 1 ) 



55 



Security Classification 



A-31408 



Security Classification 



KEY WO R OS 



electrogasdynamic 
corona discharge 
ion injector 
energy conversion 



DD ,^1473 back 



S/N 01 Ot -807-6821 



-1 



56 



Security Classification 



A - 3 I 409 



DISTRIBUTION LIST 

No. Copies 

1. Defense Documentation Center 20 
Cameron Station 

Alexandria, Virginia 22314 

2. Library 2 
Naval Postgraduate School 

Monterey, California 93940 

3. Commander 3 
Naval Air Systems Command 

Department of the Navy 

Attn: Pr. H. R. Rosenwasser, Code AIR-310C 

4. Chairman 1 
Department of Aeronautics 

Naval Postgraduate School 
Monterey, California 93940 

5. Professor T. H. Gawain 1 
Department of Aeronautics 

Naval Postgraduate School 
Monterey, California 93940 

6. Professor K. E. Woehler 1 
Physics Department 

Naval Postgraduate School 
Monterey, California 93940 

7. Professor Oscar Biblarz 10 
Department of Aeronautics 

Naval Postgraduate School 
Monterey, California 93940 

8. Professor J. R. Melcher 1 
Dept. of Elec. Engr. , Room 31-141 

Mass. Inst, of Technology 
Cambridge, Massachusetts 02139 

9. Professor Albert Solbes 1 
Dept. of Aero., Room 37-375 

Mass. Inst, of Technology 
Cambridge, Massachusetts 02139 

10. Mr. Alvin M. Marks 1 

Marks Polarized Corp. 
153-16 10th Avenue 
Whitestone, New York 11357 



4 7 



11. Dr. M. C. Gourdine 
Gourd ine Systems, Inc. 
112 Nay Ion Avenue 
Livingston, New Jersey 07039 

12. Dr. Hans von Ohain 
Aerospace Research Laboratory 
U. S. Air Force 
Wright-Patterson AFB, Ohio 45433 

13. Dr. A. E. Fuhs 

Chief Scientist -Code APX 
Aero Propulsion Laboratory 
Wright-Patterson AFB, Ohio 45433 

14. Dr. Ernesto Barreto 
Atmospheric Research Center 
State University of New York 
130 Saratoga Road 

Scotia, New York 12302 

15. Professor H. R. Velkoff 
Dept. of Mech. Engr. 
Ohio State University 
Columbus, Ohio 43210 

16. Dr. Otmar M. Stuetzer 
Sandia Corporation 

P. 0. Box 5800 

Albuquerque, New Mexico 87115 

17. Dr. Ralph Roberts 
Office of Naval Research 
Power Program, Code 473 
Washington, D. C. 20360 

18. Mr. John A. Stakowski 
Office of Naval Research 
Power Program, Code 473 
Washington, D. C. 20360 

19. Dr. H. J. Mueller 

Naval Air Systems Command 

Code AIR 310 

Washington, D. C. 20360 



<r 



20. Dr. S. J. Magram 
Army Research Office 
Arlington, Virginia 22200 

21. Air Force Office of Scientific Research 
Washington, D. C. 20333 

ATTN: Power Systems Group 

22. Mr. Robert C. Hamilton 
Institute of Defense Analysis 
400 Army - Navy Drive 
Arlington, Virginia 22202 

23. Dr. George C. Szego 
Institute of Defense Analysis 
400 Army - Navy Drive 
Arlington, Virginia 22202 

24. Mr. S. Cohen 

NASA Lewis Research Center 
21000 Brookpark Road 
Cleveland, Ohio 44135 

25. Mr. M. Lawson 

Aerospace Research Laboratory 

U. S. Air Force 

Wright-Patterson AFB, Ohio 45433 

26. Dean of Research Administration 
Naval Postgraduate School 
Monterey, California 93940 

27. LTJG William T. Ober III, USN 
76 Salem Street 

Andover, Massachusetts 01810 



oi 



Unclassified 



Security Classification 



DOCUMENT CONTROL DATA -R&D 

< Security classification ot title, body ol abstract and indexing annotation must be entered when the overall report Is classified) 



I originating activity (Corporate author) 

Naval Postgraduate School 
Monterey, California 93940 



2a. REPORT SECURITY CLASSIFICATION 



Unclassified 



2b. GROUP 



3 REPORT TITLE 



EHD Research 

Final Report for the Year 1968-69 



4 descriptive NOTES (Type ol report and, in ejus ive dates) 



5 au thoriSI (First name, middle initial, last name) 

Oscar Biblarz 



6 REPOR T D A TE 

December 1969 



7a. TOTAL NO. OF PAGES 



7b. NO. OF RE FS 

28 



8a. CONTRACT OR GRANT NO. 

AIRTASK No A34340/551/69R01002010 



0a. ORIGINATOR'S REPORT NUMBER(S) 



b. PROJEC T NO 



NPS-57ZI9121A 



9b. OTHER REPORT NO(SI (Any other numbers that may be asslonrd 
this report) " 



10 DISTRIBUTION STATEMENT 



This document has been approved for public release and sale; its distribution is 
unlimited. 



II SUPPLEMENTARY NOTES 



12. SPONSORING MILI TARY ACTIVITY 

Naval Air Systems Command 
Washington, D. C. 



13. ABSTR AC T 



The present research in electrohydrodynamics is concerned with how charged 
particles can be generated in the laboratory with a potentially useful range of sizes, 
of charge, and of number density. It is suggested that refined measurement techniques 
are needed to check on just what is being injected into the flow. The effects of 
turbulence on the EHD process and, particularly, on breakdown are being studied. The 
report discusses in some detail the possible role of turbulence on the mean effective 
mobility of charged particles. 

On the experimental side, a laboratory facility has been built and then improved 
by the addition of a larger test section and other equipment. Work is proceeding to 
further develop and refine the instrumentation. Two types of injectors have been 
operated, namely, molecular and two phase and the latter shows potential for efficient 
operation. 

It has been concluded tentatively that turbulence in the carrier fluid Increases 
its breakdown potential, and that turbulent air may be a suitable medium for the EHD 
energy conversion process. 

Research plans for the coming year are outlined in the report. 



DD 



F0R " 1473 

i nov ee I *T # «■/ 

S/N 0101 -807-681 1 



(PAGE 1) 



/' 



Unclassified 

Security Classification 



A- 3 1408 



UNCLASSIFIED 



Security Classification 



k e v wo rd s 



Elect rohydrodynamics 

Turbulence 

Breakdown potential 

Charged Particle injector 

Mobility 

Direct Energy Conversion 



DD , F r. u .,1473 'back, 



S/N 01O1 - 907-632 I 



ROLE AT 



£ <f- 



UNCLASSIFIED 



Security Classification 



A - 3 I 409 



MI«!?EL KN0X LIBRARY 




3 2768 00396357